diff --git a/.prettierignore b/.prettierignore
new file mode 100644
index 0000000..6a398ea
--- /dev/null
+++ b/.prettierignore
@@ -0,0 +1,14 @@
+# Build output and tool caches
+dist
+.netlify
+.tanstack
+.output
+src/routeTree.gen.ts
+
+# Golden files: tests compare these byte for byte
+tests/fixtures
+
+# Vendored font binaries and their licence
+assets/fonts
+
+package-lock.json
diff --git a/assets/fonts/Inter-Bold.woff b/assets/fonts/Inter-Bold.woff
new file mode 100644
index 0000000..23882e4
Binary files /dev/null and b/assets/fonts/Inter-Bold.woff differ
diff --git a/assets/fonts/Inter-Regular.woff b/assets/fonts/Inter-Regular.woff
new file mode 100644
index 0000000..df5bd3b
Binary files /dev/null and b/assets/fonts/Inter-Regular.woff differ
diff --git a/assets/fonts/OFL.txt b/assets/fonts/OFL.txt
new file mode 100644
index 0000000..21f6aff
--- /dev/null
+++ b/assets/fonts/OFL.txt
@@ -0,0 +1,93 @@
+Copyright 2020 The Inter Project Authors (https://github.com/rsms/inter)
+
+This Font Software is licensed under the SIL Open Font License, Version 1.1.
+This license is copied below, and is also available with a FAQ at:
+https://scripts.sil.org/OFL
+
+
+-----------------------------------------------------------
+SIL OPEN FONT LICENSE Version 1.1 - 26 February 2007
+-----------------------------------------------------------
+
+PREAMBLE
+The goals of the Open Font License (OFL) are to stimulate worldwide
+development of collaborative font projects, to support the font creation
+efforts of academic and linguistic communities, and to provide a free and
+open framework in which fonts may be shared and improved in partnership
+with others.
+
+The OFL allows the licensed fonts to be used, studied, modified and
+redistributed freely as long as they are not sold by themselves. The
+fonts, including any derivative works, can be bundled, embedded, 
+redistributed and/or sold with any software provided that any reserved
+names are not used by derivative works. The fonts and derivatives,
+however, cannot be released under any other type of license. The
+requirement for fonts to remain under this license does not apply
+to any document created using the fonts or their derivatives.
+
+DEFINITIONS
+"Font Software" refers to the set of files released by the Copyright
+Holder(s) under this license and clearly marked as such. This may
+include source files, build scripts and documentation.
+
+"Reserved Font Name" refers to any names specified as such after the
+copyright statement(s).
+
+"Original Version" refers to the collection of Font Software components as
+distributed by the Copyright Holder(s).
+
+"Modified Version" refers to any derivative made by adding to, deleting,
+or substituting -- in part or in whole -- any of the components of the
+Original Version, by changing formats or by porting the Font Software to a
+new environment.
+
+"Author" refers to any designer, engineer, programmer, technical
+writer or other person who contributed to the Font Software.
+
+PERMISSION & CONDITIONS
+Permission is hereby granted, free of charge, to any person obtaining
+a copy of the Font Software, to use, study, copy, merge, embed, modify,
+redistribute, and sell modified and unmodified copies of the Font
+Software, subject to the following conditions:
+
+1) Neither the Font Software nor any of its individual components,
+in Original or Modified Versions, may be sold by itself.
+
+2) Original or Modified Versions of the Font Software may be bundled,
+redistributed and/or sold with any software, provided that each copy
+contains the above copyright notice and this license. These can be
+included either as stand-alone text files, human-readable headers or
+in the appropriate machine-readable metadata fields within text or
+binary files as long as those fields can be easily viewed by the user.
+
+3) No Modified Version of the Font Software may use the Reserved Font
+Name(s) unless explicit written permission is granted by the corresponding
+Copyright Holder. This restriction only applies to the primary font name as
+presented to the users.
+
+4) The name(s) of the Copyright Holder(s) or the Author(s) of the Font
+Software shall not be used to promote, endorse or advertise any
+Modified Version, except to acknowledge the contribution(s) of the
+Copyright Holder(s) and the Author(s) or with their explicit written
+permission.
+
+5) The Font Software, modified or unmodified, in part or in whole,
+must be distributed entirely under this license, and must not be
+distributed under any other license. The requirement for fonts to
+remain under this license does not apply to any document created
+using the Font Software.
+
+TERMINATION
+This license becomes null and void if any of the above conditions are
+not met.
+
+DISCLAIMER
+THE FONT SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT
+OF COPYRIGHT, PATENT, TRADEMARK, OR OTHER RIGHT. IN NO EVENT SHALL THE
+COPYRIGHT HOLDER BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
+INCLUDING ANY GENERAL, SPECIAL, INDIRECT, INCIDENTAL, OR CONSEQUENTIAL
+DAMAGES, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+FROM, OUT OF THE USE OR INABILITY TO USE THE FONT SOFTWARE OR FROM
+OTHER DEALINGS IN THE FONT SOFTWARE.
diff --git a/content/episodes/2024-01.md b/content/episodes/2024-01.md
index fd58589..9928ad2 100644
--- a/content/episodes/2024-01.md
+++ b/content/episodes/2024-01.md
@@ -161,7 +161,7 @@ And that combined with TypeScript really taking over as the definitive way to wr
 
 And then maybe some of the learnings from running a language like that will be brought back into JavaScript, the ECMAScript process, or into TypeScript itself. Love to see it. [00:07:56]
 
- [00:08:07]
+[00:08:07]
 
 ## [React Libraries for 2024 by Robin Wieruch](https://www.robinwieruch.de/react-libraries/)
 
@@ -288,7 +288,7 @@ And very cool to have your source maps PR ship out. Finally! Ha ha ha. Looking f
 
 **Carl Vitullo:** Well, hopefully it'll make the stack traces easier to read anyway. [00:22:30]
 
- [00:22:33]
+[00:22:33]
 
 ## [Remix SPA mode](https://github.com/remix-run/remix/pull/8338)
 
diff --git a/content/episodes/2024-06.md b/content/episodes/2024-06.md
index b5fc7a4..534ced5 100644
--- a/content/episodes/2024-06.md
+++ b/content/episodes/2024-06.md
@@ -629,7 +629,7 @@ Lock will let you pin your dependencies, left the, you know, the gap of what abo
 
 So being able to pin a version, super great. Yeah. As well as like having a env loader, so we don't need to use env anymore. But yeah, so I appreciated this as a quick summary of a number of new built ins for Node. [00:57:06]
 
- [00:57:18]
+[00:57:18]
 
 ## [Migrating to the Next App Router With Zero Downtime](https://workos.com/blog/migrating-to-next-js-app-router-with-zero-downtime)
 
diff --git a/content/episodes/2024-07.md b/content/episodes/2024-07.md
index e983113..7aab023 100644
--- a/content/episodes/2024-07.md
+++ b/content/episodes/2024-07.md
@@ -137,7 +137,7 @@ Yeah, speaking of layoffs, layoffs. fyi. This is the first time in, like, the A
 
 **Carl Vitullo:** I've never quite understood the niche, the purpose that Storybook really serves. But, you know, hey, minor release, so it seemed worth calling out. I know it's pretty widely used. [00:04:42]
 
- [00:04:52]
+[00:04:52]
 
 ## [React Native Reanimated 3.14](https://github.com/software-mansion/)
 
@@ -163,7 +163,7 @@ So this is a Rust based developer being built by the team that put together one
 
 **Mark Erikson:** React Rally is wonderful. They've done a great job of really having a good community vibe over the years. I should be there. So folks want to come by and say hi, please do. [00:06:16]
 
- [00:06:24]
+[00:06:24]
 
 ## [React Universe](https://www.reactuniverseconf.com/)
 
diff --git a/content/episodes/2025-02.md b/content/episodes/2025-02.md
index 352ba57..d956b41 100644
--- a/content/episodes/2025-02.md
+++ b/content/episodes/2025-02.md
@@ -167,7 +167,7 @@ I've done a lot of maintenance of legacy apps and weird edge cases like that, li
 
 ## [Prettier 3.5](https://prettier.io/blog/2025/02/09/3.5.0)
 
- [00:09:01]
+[00:09:01]
 
 **Carl Vitullo:** Cool, new version of Prettier. I saw they're adding an option, which they do not do very frequently, so that's cool. [00:09:01]
 
@@ -455,7 +455,7 @@ And so Ben's pretty excited about that. Also on, also on the note of just, you k
 
 ## [JavaScript Temporal is coming](https://developer.mozilla.org/en-US/blog/javascript-temporal-is-coming/)
 
- [00:49:52]
+[00:49:52]
 
 **Carl Vitullo:** Mozilla had a post at the end of January, about Temporal. JavaScript Temporal is coming. Dates in JavaScript are garbage. They are awful. My favorite example of a specific way they're awful is that they have no concept of time zone. And so every date uses your local system time zone, Which is absolutely awful. [00:49:53]
 
diff --git a/docs/superpowers/plans/2026-09-08-content-pipeline.md b/docs/superpowers/plans/2026-09-08-content-pipeline.md
index 92d8183..d7c4884 100644
--- a/docs/superpowers/plans/2026-09-08-content-pipeline.md
+++ b/docs/superpowers/plans/2026-09-08-content-pipeline.md
@@ -27,22 +27,22 @@
 
 ## File Structure
 
-| Path | Responsibility |
-| --- | --- |
-| `package.json` | `type: module`, `private`, `test` script, deps |
-| `tsconfig.json` | Editor config only (`noEmit`, `allowImportingTsExtensions`, `erasableSyntaxOnly`) |
-| `.gitignore` | `node_modules`, `.DS_Store`, `*.m4a` |
-| `src/content/slug.ts` | `slug()`, `normalizeTime()`, `flattenLinks()` — the shared text rules |
-| `src/content/parse.ts` | Types (`Episode`, `OutlineItem`, `Section`, `Segment`), `splitFile()`, `parseEpisode()` |
-| `src/content/serialize.ts` | `serializeEpisodeFile()` — front matter + untouched body back to a string |
-| `src/content/srt.ts` | `toSrt(episode)` |
-| `scripts/migrate.ts` | One-time conversion of `reactiflux.com/src/transcripts/tmir-*.md` |
-| `scripts/ingest.ts` | Transistor feed → ingest-owned front matter |
-| `scripts/publish-transcript.ts` | Descript export → file body; SRT → Transistor |
-| `scripts/publish-atproto.ts` | `site.standard.publication` + `site.standard.document` records → the show's PDS; `atUri` → front matter |
-| `.env` (gitignored), `.env.example` | every credential and build variable, loaded by `node --env-file-if-exists=.env` |
-| `tests/*.test.ts` | One test file per module/script |
-| `tests/fixtures/*.md`, `tests/fixtures/feed.xml` | Short real excerpts of each historical format |
+| Path                                             | Responsibility                                                                                          |
+| ------------------------------------------------ | ------------------------------------------------------------------------------------------------------- |
+| `package.json`                                   | `type: module`, `private`, `test` script, deps                                                          |
+| `tsconfig.json`                                  | Editor config only (`noEmit`, `allowImportingTsExtensions`, `erasableSyntaxOnly`)                       |
+| `.gitignore`                                     | `node_modules`, `.DS_Store`, `*.m4a`                                                                    |
+| `src/content/slug.ts`                            | `slug()`, `normalizeTime()`, `flattenLinks()` — the shared text rules                                   |
+| `src/content/parse.ts`                           | Types (`Episode`, `OutlineItem`, `Section`, `Segment`), `splitFile()`, `parseEpisode()`                 |
+| `src/content/serialize.ts`                       | `serializeEpisodeFile()` — front matter + untouched body back to a string                               |
+| `src/content/srt.ts`                             | `toSrt(episode)`                                                                                        |
+| `scripts/migrate.ts`                             | One-time conversion of `reactiflux.com/src/transcripts/tmir-*.md`                                       |
+| `scripts/ingest.ts`                              | Transistor feed → ingest-owned front matter                                                             |
+| `scripts/publish-transcript.ts`                  | Descript export → file body; SRT → Transistor                                                           |
+| `scripts/publish-atproto.ts`                     | `site.standard.publication` + `site.standard.document` records → the show's PDS; `atUri` → front matter |
+| `.env` (gitignored), `.env.example`              | every credential and build variable, loaded by `node --env-file-if-exists=.env`                         |
+| `tests/*.test.ts`                                | One test file per module/script                                                                         |
+| `tests/fixtures/*.md`, `tests/fixtures/feed.xml` | Short real excerpts of each historical format                                                           |
 
 ## Canonical Episode File Format
 
@@ -99,6 +99,7 @@ Rules:
 ### Task 1: Repo bootstrap and the shared text rules
 
 **Files:**
+
 - Create: `package.json`
 - Create: `tsconfig.json`
 - Create: `.gitignore`
@@ -106,6 +107,7 @@ Rules:
 - Test: `tests/slug.test.ts`
 
 **Interfaces:**
+
 - Consumes: nothing.
 - Produces:
   - `slug(title: string): string`
@@ -205,7 +207,10 @@ test("slug matches anchors published in existing transcripts", () => {
 });
 
 test("slug keeps underscores and digits, drops everything else non-word", () => {
-  assert.equal(slug("React Strict DOM, Why is it so Great?"), "react-strict-dom-why-is-it-so-great");
+  assert.equal(
+    slug("React Strict DOM, Why is it so Great?"),
+    "react-strict-dom-why-is-it-so-great",
+  );
   assert.equal(slug("Node.js v22"), "nodejs-v22");
 });
 
@@ -292,32 +297,70 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 ### Task 2: Episode parser and serializer
 
 **Files:**
+
 - Create: `src/content/parse.ts`
 - Create: `src/content/serialize.ts`
 - Create: `tests/fixtures/canonical-2026-05.md`
 - Test: `tests/parse.test.ts`
 
 **Interfaces:**
+
 - Consumes: `slug`, `flattenLinks`, `normalizeTime` from `src/content/slug.ts`.
 - Produces:
+
   ```ts
-  export interface Person { name: string; role?: string; href?: string; img?: string }
-  export interface OutlineItem { title: string; url?: string; time?: string; anchor: string; children: OutlineItem[] }
-  export interface Segment { speaker: string; time?: string; text: string }
-  export interface Section { title: string; anchor: string; time?: string; segments: Segment[] }
+  export interface Person {
+    name: string;
+    role?: string;
+    href?: string;
+    img?: string;
+  }
+  export interface OutlineItem {
+    title: string;
+    url?: string;
+    time?: string;
+    anchor: string;
+    children: OutlineItem[];
+  }
+  export interface Segment {
+    speaker: string;
+    time?: string;
+    text: string;
+  }
+  export interface Section {
+    title: string;
+    anchor: string;
+    time?: string;
+    segments: Segment[];
+  }
   export interface Episode {
-    slug: string; title: string; date: string; description: string;
-    time?: string; location?: string;
-    transistorId?: string; audioUrl?: string; duration?: number;
-    season?: number; episode?: number; people: Person[];
-    bskyPostUrl?: string;   // written by ingest (Task 4)
-    atUri?: string;         // written by publish-atproto (Task 7), never by ingest
-    outline: OutlineItem[]; sections: Section[];
+    slug: string;
+    title: string;
+    date: string;
+    description: string;
+    time?: string;
+    location?: string;
+    transistorId?: string;
+    audioUrl?: string;
+    duration?: number;
+    season?: number;
+    episode?: number;
+    people: Person[];
+    bskyPostUrl?: string; // written by ingest (Task 4)
+    atUri?: string; // written by publish-atproto (Task 7), never by ingest
+    outline: OutlineItem[];
+    sections: Section[];
   }
-  export function splitFile(text: string): { frontMatter: Record<string, unknown>; body: string }
-  export function parseOutline(region: string): OutlineItem[]
-  export function parseEpisode(text: string, epSlug: string): Episode
-  export function serializeEpisodeFile(frontMatter: Record<string, unknown>, body: string): string  // from serialize.ts
+  export function splitFile(text: string): {
+    frontMatter: Record<string, unknown>;
+    body: string;
+  };
+  export function parseOutline(region: string): OutlineItem[];
+  export function parseEpisode(text: string, epSlug: string): Episode;
+  export function serializeEpisodeFile(
+    frontMatter: Record<string, unknown>,
+    body: string,
+  ): string; // from serialize.ts
   ```
 
 - [ ] **Step 1: Create the canonical fixture**
@@ -451,7 +494,10 @@ test("speaker carries forward across consecutive paragraphs", () => {
   assert.equal(segs[0].speaker, "Carl Vitullo");
   assert.equal(segs[0].time, "00:00:55");
   assert.equal(segs[1].speaker, "Carl Vitullo");
-  assert.equal(segs[1].text, "Stuff has changed. Stuff will continue to change. Just like life.");
+  assert.equal(
+    segs[1].text,
+    "Stuff has changed. Stuff will continue to change. Just like life.",
+  );
   assert.equal(segs[2].speaker, "Mark Erikson");
 });
 
@@ -465,7 +511,11 @@ test("serialize writes a changed field without disturbing the body", () => {
   frontMatter.duration = 5000;
   const out = serializeEpisodeFile(frontMatter, body);
   assert.match(out, /^duration: 5000$/m);
-  assert.ok(out.includes("**Mark Erikson:** Yeah, that sounds all too real. [00:01:44]"));
+  assert.ok(
+    out.includes(
+      "**Mark Erikson:** Yeah, that sounds all too real. [00:01:44]",
+    ),
+  );
 });
 ```
 
@@ -715,6 +765,7 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 ### Task 3: Migrate historical transcripts
 
 **Files:**
+
 - Create: `scripts/migrate.ts`
 - Create: `tests/fixtures/legacy-2023-march.md`
 - Create: `tests/fixtures/legacy-2024-03.md`
@@ -723,9 +774,10 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 - Create: `tests/fixtures/legacy-2026-05.md`
 - Test: `tests/migrate.test.ts`
 
-**Where the transcript starts.** The boundary is the first *speaker paragraph* — a block matching `^(\[\d…\]\s*)?\*\*[^*]+:\*\*`. But a `##` section heading can sit directly above that paragraph (as in `tmir-2024-03.md`, whose transcript opens with `## Layoffs (seems better!)` before the first `**Carl Vitullo:**` block in some months), so after finding the first speaker paragraph, walk *backwards* over any contiguous `##` heading blocks and include them in the transcript. Walk back over `##` only, never `#`: a lone `#` heading immediately above the transcript is the old transcript-start marker (`# Interview`, `# This Month in React: June, 2025`, `# This Month in React 2023 April`) and is dropped silently, while an `#` heading anywhere else in the outline region (`# Main Content` in `tmir-2024-04.md`) is dropped with a warning.
+**Where the transcript starts.** The boundary is the first _speaker paragraph_ — a block matching `^(\[\d…\]\s*)?\*\*[^*]+:\*\*`. But a `##` section heading can sit directly above that paragraph (as in `tmir-2024-03.md`, whose transcript opens with `## Layoffs (seems better!)` before the first `**Carl Vitullo:**` block in some months), so after finding the first speaker paragraph, walk _backwards_ over any contiguous `##` heading blocks and include them in the transcript. Walk back over `##` only, never `#`: a lone `#` heading immediately above the transcript is the old transcript-start marker (`# Interview`, `# This Month in React: June, 2025`, `# This Month in React 2023 April`) and is dropped silently, while an `#` heading anywhere else in the outline region (`# Main Content` in `tmir-2024-04.md`) is dropped with a warning.
 
 **Interfaces:**
+
 - Consumes: `normalizeTime`, `slug`, `flattenLinks` from `src/content/slug.ts`; `parseEpisode` from `src/content/parse.ts`; `serializeEpisodeFile` from `src/content/serialize.ts`.
 - Produces:
   - `slugFromFilename(name: string): string | null` — `"tmir-march-2023.md"` → `"2023-03"`, `"tmir-2024-03.md"` → `"2024-03"`
@@ -735,15 +787,15 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 
 There is no clean three-era split — the markers vary per file. Detect each feature independently rather than branching on an era:
 
-| Feature | Variants observed |
-| --- | --- |
-| Header block | `<iframe …>` (36 files); `<style>…</style>` before it from 2025-01 on; three files have no iframe at all (`tmir-2025-09.md`, `tmir-dec-2023.md`, `tmir-nov-2023.md`) |
-| Pre-outline link list | `- [Spotify](…)` / `- [Apple Podcasts](…)` / `- [RSS](…)` bullets before the iframe (2023–2024) |
-| Outline | absent (all of 2023 except dec/nov, plus `tmir-2025-06.md`); `- [[mm:ss](#anchor)] …`; `- [[hh:mm:ss](#anchor)] …` (2026) |
-| Transcript start | no marker at all; or an h1 like `# This Month in React: June, 2025` / `# Interview` / `# This Month in React 2023 April` |
-| Section headings | `## Title` (most); `# Title` (`tmir-june-2023.md`, five of them); `## [00:52] title` (`tmir-2025-06.md`) |
-| Paragraph timestamp | trailing `… [00:16]`; leading `[01:23] **Carl Vitullo:** …` |
-| Speaker labels | `**Carl Vitullo:**`, `**Carl:**`, raw Descript ids `**1-vcarl:**` (`tmir-2025-06.md`) |
+| Feature               | Variants observed                                                                                                                                                    |
+| --------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| Header block          | `<iframe …>` (36 files); `<style>…</style>` before it from 2025-01 on; three files have no iframe at all (`tmir-2025-09.md`, `tmir-dec-2023.md`, `tmir-nov-2023.md`) |
+| Pre-outline link list | `- [Spotify](…)` / `- [Apple Podcasts](…)` / `- [RSS](…)` bullets before the iframe (2023–2024)                                                                      |
+| Outline               | absent (all of 2023 except dec/nov, plus `tmir-2025-06.md`); `- [[mm:ss](#anchor)] …`; `- [[hh:mm:ss](#anchor)] …` (2026)                                            |
+| Transcript start      | no marker at all; or an h1 like `# This Month in React: June, 2025` / `# Interview` / `# This Month in React 2023 April`                                             |
+| Section headings      | `## Title` (most); `# Title` (`tmir-june-2023.md`, five of them); `## [00:52] title` (`tmir-2025-06.md`)                                                             |
+| Paragraph timestamp   | trailing `… [00:16]`; leading `[01:23] **Carl Vitullo:** …`                                                                                                          |
+| Speaker labels        | `**Carl Vitullo:**`, `**Carl:**`, raw Descript ids `**1-vcarl:**` (`tmir-2025-06.md`)                                                                                |
 
 Two traps confirmed by reading the files:
 
@@ -984,7 +1036,9 @@ test("2023 format: iframe stripped, outline synthesized from headings", () => {
   );
   assert.ok(text.includes("\n# Transcript\n"));
   assert.ok(
-    text.includes("**Carl Vitullo:** Thanks everyone for joining us for this month in React. [00:00:16]"),
+    text.includes(
+      "**Carl Vitullo:** Thanks everyone for joining us for this month in React. [00:00:16]",
+    ),
   );
   assert.deepEqual(warnings, []);
 
@@ -1000,13 +1054,19 @@ test("2024 format: link list dropped, leading timestamps moved to the end", () =
   assert.ok(!text.includes("[Spotify]("));
   assert.ok(!text.includes("<iframe"));
   assert.ok(
-    text.includes("- Quick Hits\n  - [[00:00:39](#layoffsfyi)] [Layoffs.fyi](https://layoffs.fyi/)\n"),
+    text.includes(
+      "- Quick Hits\n  - [[00:00:39](#layoffsfyi)] [Layoffs.fyi](https://layoffs.fyi/)\n",
+    ),
     text,
   );
   assert.ok(
-    text.includes("**Carl Vitullo:** I'm going to start off with some layoffs news. [00:00:39]"),
+    text.includes(
+      "**Carl Vitullo:** I'm going to start off with some layoffs news. [00:00:39]",
+    ),
+  );
+  assert.ok(
+    text.includes("This month has about 7, 200 laid off so far. [00:00:58]"),
   );
-  assert.ok(text.includes("This month has about 7, 200 laid off so far. [00:00:58]"));
 
   const ep = parseEpisode(text, "2024-03");
   assert.equal(ep.sections.length, 3); // Intro, Layoffs, New releases
@@ -1097,8 +1157,18 @@ import { splitFile } from "../src/content/parse.ts";
 import { serializeEpisodeFile } from "../src/content/serialize.ts";
 
 const MONTHS = [
-  "january", "february", "march", "april", "may", "june",
-  "july", "august", "september", "october", "november", "december",
+  "january",
+  "february",
+  "march",
+  "april",
+  "may",
+  "june",
+  "july",
+  "august",
+  "september",
+  "october",
+  "november",
+  "december",
 ];
 
 /** "tmir-2024-03.md" | "tmir-march-2023.md" | "tmir-dec-2023.md" -> "yyyy-mm" */
@@ -1131,7 +1201,10 @@ export function convert(raw: string): { text: string; warnings: string[] } {
   const warnings: string[] = [];
   const { frontMatter, body } = splitFile(raw);
   const clean = stripEmbeds(body);
-  const blocks = clean.split(/\n{2,}/).map((b) => b.trim()).filter(Boolean);
+  const blocks = clean
+    .split(/\n{2,}/)
+    .map((b) => b.trim())
+    .filter(Boolean);
 
   // The transcript starts at the first speaker paragraph, plus any `##`
   // section headings sitting directly above it. Walk back over `##` only:
@@ -1140,7 +1213,9 @@ export function convert(raw: string): { text: string; warnings: string[] } {
   // `# Main Content`).
   let start = blocks.findIndex((b) => SPEAKER_PARA.test(b));
   if (start === -1) {
-    warnings.push("no speaker paragraphs found; body left in the outline region");
+    warnings.push(
+      "no speaker paragraphs found; body left in the outline region",
+    );
   }
   while (start > 0 && blocks[start - 1].startsWith("## ")) start -= 1;
 
@@ -1153,7 +1228,10 @@ export function convert(raw: string): { text: string; warnings: string[] } {
     if (!OUTLINE_LINE.test(block.split("\n")[0])) {
       const isMarker = i === head.length - 1 && /^#\s/.test(block);
       if (isMarker) continue; // old transcript-start marker; drop silently
-      if (block.startsWith("#")) warnings.push(`dropped heading before outline: ${block.split("\n")[0]}`);
+      if (block.startsWith("#"))
+        warnings.push(
+          `dropped heading before outline: ${block.split("\n")[0]}`,
+        );
       else warnings.push(`dropped prose before outline: ${block.slice(0, 60)}`);
       continue;
     }
@@ -1162,7 +1240,10 @@ export function convert(raw: string): { text: string; warnings: string[] } {
       if (!m) continue;
       let rest = m[2].trim();
       // Drop the pre-iframe subscribe link list.
-      if (/^\[(Spotify|Apple Podcasts|RSS|or anywhere you prefer)\]\(/.test(rest)) continue;
+      if (
+        /^\[(Spotify|Apple Podcasts|RSS|or anywhere you prefer)\]\(/.test(rest)
+      )
+        continue;
 
       let time: string | undefined;
       const t = OUTLINE_TIME.exec(rest);
@@ -1200,7 +1281,8 @@ export function convert(raw: string): { text: string; warnings: string[] } {
       // 2023 used h1 for sections; promote to h2.
       pendingHeadingAnchor = slug(title);
       out.push(`## ${title}`);
-      if (level > 2) warnings.push(`heading deeper than h2 flattened: ${title}`);
+      if (level > 2)
+        warnings.push(`heading deeper than h2 flattened: ${title}`);
       continue;
     }
 
@@ -1218,11 +1300,16 @@ export function convert(raw: string): { text: string; warnings: string[] } {
         text = text.slice(0, trail.index);
       }
     }
-    if (!time) warnings.push(`paragraph without timestamp: ${text.slice(0, 60)}`);
+    if (!time)
+      warnings.push(`paragraph without timestamp: ${text.slice(0, 60)}`);
     if (!/^\*\*[^*]+:\*\*/.test(text) && out.length === 0) {
       warnings.push(`paragraph without speaker: ${text.slice(0, 60)}`);
     }
-    if (pendingHeadingAnchor && time && !sectionTimes.has(pendingHeadingAnchor)) {
+    if (
+      pendingHeadingAnchor &&
+      time &&
+      !sectionTimes.has(pendingHeadingAnchor)
+    ) {
       sectionTimes.set(pendingHeadingAnchor, time);
       pendingHeadingAnchor = null;
     }
@@ -1236,7 +1323,9 @@ export function convert(raw: string): { text: string; warnings: string[] } {
       const title = line.slice(3).trim();
       const anchor = slug(title);
       const time = sectionTimes.get(anchor);
-      outlineLines.push(time ? `- [[${time}](#${anchor})] ${title}` : `- ${title}`);
+      outlineLines.push(
+        time ? `- [[${time}](#${anchor})] ${title}` : `- ${title}`,
+      );
     }
   }
 
@@ -1257,10 +1346,14 @@ function main() {
   for (const name of readdirSync(SOURCE_DIR).sort()) {
     const epSlug = slugFromFilename(name);
     if (!epSlug) continue;
-    const { text, warnings } = convert(readFileSync(join(SOURCE_DIR, name), "utf8"));
+    const { text, warnings } = convert(
+      readFileSync(join(SOURCE_DIR, name), "utf8"),
+    );
     writeFileSync(join(OUT_DIR, `${epSlug}.md`), text);
     files += 1;
-    console.log(`${name} -> content/episodes/${epSlug}.md  (${warnings.length} warnings)`);
+    console.log(
+      `${name} -> content/episodes/${epSlug}.md  (${warnings.length} warnings)`,
+    );
     for (const w of warnings) {
       console.log(`    ${w}`);
       warned += 1;
@@ -1307,11 +1400,13 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 ### Task 4: Ingest Transistor feed metadata
 
 **Files:**
+
 - Create: `scripts/ingest.ts`
 - Create: `tests/fixtures/feed.xml`
 - Test: `tests/ingest.test.ts`
 
 **Interfaces:**
+
 - Consumes: `splitFile` from `src/content/parse.ts`; `serializeEpisodeFile` from `src/content/serialize.ts`.
 - Produces:
   - `interface FeedItem { slug: string; transistorId: string; audioUrl: string; duration: number; season?: number; episode?: number; people: Person[]; bskyPostUrl?: string }`
@@ -1402,17 +1497,35 @@ import { readFileSync } from "node:fs";
 import { parseFeed, slugFromTitle, applyFeedItem } from "../scripts/ingest.ts";
 import { parseEpisode } from "../src/content/parse.ts";
 
-const xml = readFileSync(new URL("./fixtures/feed.xml", import.meta.url), "utf8");
+const xml = readFileSync(
+  new URL("./fixtures/feed.xml", import.meta.url),
+  "utf8",
+);
 
 test("slugFromTitle handles every TMiR title style and rejects the rest", () => {
-  assert.equal(slugFromTitle("TMiR 2026-05: Who even is on the Core team anymore"), "2026-05");
-  assert.equal(slugFromTitle("TMiR 2023-10: React Forget, Canary Releases"), "2023-10");
-  assert.equal(slugFromTitle("This Month in React – September 2023"), "2023-09");
+  assert.equal(
+    slugFromTitle("TMiR 2026-05: Who even is on the Core team anymore"),
+    "2026-05",
+  );
+  assert.equal(
+    slugFromTitle("TMiR 2023-10: React Forget, Canary Releases"),
+    "2023-10",
+  );
+  assert.equal(
+    slugFromTitle("This Month in React – September 2023"),
+    "2023-09",
+  );
   assert.equal(slugFromTitle("This Month In React – March 2023"), "2023-03");
   assert.equal(slugFromTitle("This Month in React (April 2023)"), "2023-04");
-  assert.equal(slugFromTitle("Office Hours – States of Burnout with Jenny Truong"), null);
+  assert.equal(
+    slugFromTitle("Office Hours – States of Burnout with Jenny Truong"),
+    null,
+  );
   assert.equal(slugFromTitle("Behind the React Documentary"), null);
-  assert.equal(slugFromTitle("Mark & Carl talk with Swizec Teller about using AI at work"), null);
+  assert.equal(
+    slugFromTitle("Mark & Carl talk with Swizec Teller about using AI at work"),
+    null,
+  );
 });
 
 test("parseFeed extracts TMiR items only", () => {
@@ -1439,7 +1552,8 @@ test("parseFeed extracts TMiR items only", () => {
         img: "https://img.transistorcdn.com/carl.jpg",
       },
     ],
-    bskyPostUrl: "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
+    bskyPostUrl:
+      "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
   });
   assert.equal(items[1].slug, "2023-09");
   assert.equal(items[1].people.length, 1);
@@ -1516,14 +1630,26 @@ export interface FeedItem {
 }
 
 const MONTHS = [
-  "january", "february", "march", "april", "may", "june",
-  "july", "august", "september", "october", "november", "december",
+  "january",
+  "february",
+  "march",
+  "april",
+  "may",
+  "june",
+  "july",
+  "august",
+  "september",
+  "october",
+  "november",
+  "december",
 ];
 
 function decode(s: string): string {
   return s
     .replace(/&#(\d+);/g, (_, n) => String.fromCodePoint(Number(n)))
-    .replace(/&#x([0-9a-f]+);/gi, (_, n) => String.fromCodePoint(parseInt(n, 16)))
+    .replace(/&#x([0-9a-f]+);/gi, (_, n) =>
+      String.fromCodePoint(parseInt(n, 16)),
+    )
     .replace(/&lt;/g, "<")
     .replace(/&gt;/g, ">")
     .replace(/&quot;/g, '"')
@@ -1589,9 +1715,10 @@ export function parseFeed(xml: string): FeedItem[] {
     const episode = Number(tag(item, "podcast:episode") ?? NaN);
     // The "Reply on Bluesky" anchor inside the <description> CDATA. Matched on
     // the URL shape, not the link text, so a renamed link still resolves.
-    const bskyPostUrl = /href="(https:\/\/bsky\.app\/profile\/[^/"]+\/post\/[^"]+)"/.exec(
-      item,
-    )?.[1];
+    const bskyPostUrl =
+      /href="(https:\/\/bsky\.app\/profile\/[^/"]+\/post\/[^"]+)"/.exec(
+        item,
+      )?.[1];
     items.push({
       slug: epSlug,
       transistorId,
@@ -1619,7 +1746,8 @@ export function applyFeedItem(fileText: string, item: FeedItem): string {
   if (item.season !== undefined) frontMatter.season = item.season;
   if (item.episode !== undefined) frontMatter.episode = item.episode;
   frontMatter.people = item.people;
-  if (item.bskyPostUrl !== undefined) frontMatter.bskyPostUrl = item.bskyPostUrl;
+  if (item.bskyPostUrl !== undefined)
+    frontMatter.bskyPostUrl = item.bskyPostUrl;
   return serializeEpisodeFile(frontMatter, body);
 }
 
@@ -1691,10 +1819,12 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 ### Task 5: SRT exporter
 
 **Files:**
+
 - Create: `src/content/srt.ts`
 - Test: `tests/srt.test.ts`
 
 **Interfaces:**
+
 - Consumes: `Episode` from `src/content/parse.ts`; `timestampToSeconds`, `secondsToTimestamp` from `src/content/slug.ts`.
 - Produces: `toSrt(episode: Episode): string`
 
@@ -1747,7 +1877,11 @@ test("toSrt numbers cues, formats times, and prefixes the speaker", () => {
 
 test("the last cue runs five seconds", () => {
   const srt = toSrt(parseEpisode(raw, "2026-05"));
-  assert.ok(srt.trimEnd().endsWith("Carl Vitullo: But yeah, okay, into some new releases."));
+  assert.ok(
+    srt
+      .trimEnd()
+      .endsWith("Carl Vitullo: But yeah, okay, into some new releases."),
+  );
   assert.ok(srt.includes("00:01:49,000 --> 00:01:54,000"));
 });
 ```
@@ -1776,7 +1910,9 @@ export function toSrt(episode: Episode): string {
     .filter((segment) => segment.time)
     .map((segment) => ({
       start: timestampToSeconds(segment.time!),
-      text: segment.speaker ? `${segment.speaker}: ${segment.text}` : segment.text,
+      text: segment.speaker
+        ? `${segment.speaker}: ${segment.text}`
+        : segment.text,
     }));
 
   return cues
@@ -1807,10 +1943,12 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 ### Task 6: publish-transcript script
 
 **Files:**
+
 - Create: `scripts/publish-transcript.ts`
 - Test: `tests/publish-transcript.test.ts`
 
 **Interfaces:**
+
 - Consumes: `normalizeTime` from `src/content/slug.ts`; `splitFile`, `parseEpisode`, `TRANSCRIPT_MARKER` from `src/content/parse.ts`; `serializeEpisodeFile` from `src/content/serialize.ts`; `toSrt` from `src/content/srt.ts`.
 - Produces:
   - `SPEAKERS: Record<string, string>`
@@ -1915,7 +2053,9 @@ test("replaceTranscript keeps front matter and outline, replaces the body below
   );
 
   assert.ok(!out.includes("stale text"));
-  assert.ok(out.includes("- [[00:00:55](#some-podcast-meta)] Some podcast meta"));
+  assert.ok(
+    out.includes("- [[00:00:55](#some-podcast-meta)] Some podcast meta"),
+  );
   assert.ok(out.includes("transistorId: dd8e79de"));
 
   const ep = parseEpisode(out, "2026-05");
@@ -1985,9 +2125,13 @@ export function descriptToCanonical(markdown: string): string {
 }
 
 /** Swap everything below `# Transcript`, leaving front matter and outline alone. */
-export function replaceTranscript(fileText: string, transcriptBody: string): string {
+export function replaceTranscript(
+  fileText: string,
+  transcriptBody: string,
+): string {
   const marker = fileText.indexOf(`\n${TRANSCRIPT_MARKER}\n`);
-  if (marker === -1) throw new Error(`file has no "${TRANSCRIPT_MARKER}" heading`);
+  if (marker === -1)
+    throw new Error(`file has no "${TRANSCRIPT_MARKER}" heading`);
   const head = fileText.slice(0, marker + TRANSCRIPT_MARKER.length + 1);
   return `${head}\n${transcriptBody.trimEnd()}\n`;
 }
@@ -2011,7 +2155,9 @@ export async function fetchDescriptTranscript(
     }),
   });
   if (!res.ok) {
-    throw new Error(`Descript export failed: ${res.status} ${await res.text()}`);
+    throw new Error(
+      `Descript export failed: ${res.status} ${await res.text()}`,
+    );
   }
   return res.text();
 }
@@ -2024,7 +2170,10 @@ async function findTransistorEpisodeId(
   for (let page = 1; page <= 20; page += 1) {
     const url = `${TRANSISTOR_BASE}/v1/episodes?show_id=${encodeURIComponent(showId)}&pagination[page]=${page}&pagination[per]=50`;
     const res = await fetch(url, { headers: { "x-api-key": apiKey } });
-    if (!res.ok) throw new Error(`Transistor list failed: ${res.status} ${await res.text()}`);
+    if (!res.ok)
+      throw new Error(
+        `Transistor list failed: ${res.status} ${await res.text()}`,
+      );
     const body = (await res.json()) as {
       data: { id: string; attributes: { share_url?: string } }[];
     };
@@ -2050,7 +2199,9 @@ export async function pushSrtToTranscript(
     body: JSON.stringify({ episode: { transcript_text: srt } }),
   });
   if (!res.ok) {
-    throw new Error(`Transistor update failed: ${res.status} ${await res.text()}`);
+    throw new Error(
+      `Transistor update failed: ${res.status} ${await res.text()}`,
+    );
   }
 }
 
@@ -2077,7 +2228,10 @@ async function main() {
   let fileText = readFileSync(path, "utf8");
 
   if (!skipDescript) {
-    const markdown = await fetchDescriptTranscript(projectId, requireEnv("DESCRIPT_TOKEN"));
+    const markdown = await fetchDescriptTranscript(
+      projectId,
+      requireEnv("DESCRIPT_TOKEN"),
+    );
     fileText = replaceTranscript(fileText, descriptToCanonical(markdown));
     writeFileSync(path, fileText);
     console.log(`wrote transcript into ${path}`);
@@ -2095,7 +2249,9 @@ async function main() {
       requireEnv("TRANSISTOR_API_KEY"),
       requireEnv("TRANSISTOR_SHOW_ID"),
     );
-    console.log(`pushed ${srt.split("\n\n").length} SRT cues to Transistor episode ${episode.transistorId}`);
+    console.log(
+      `pushed ${srt.split("\n\n").length} SRT cues to Transistor episode ${episode.transistorId}`,
+    );
   }
 }
 
@@ -2133,12 +2289,14 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 ### Task 7: publish-atproto script, and one `.env` for every script
 
 **Files:**
+
 - Create: `scripts/publish-atproto.ts`
 - Create: `.env.example`
 - Modify: `.gitignore`, `package.json`
 - Test: `tests/publish-atproto.test.ts`
 
 **Interfaces:**
+
 - Consumes: `Episode`, `OutlineItem`, `Person`, `parseEpisode`, `splitFile` from `src/content/parse.ts`; `serializeEpisodeFile` from `src/content/serialize.ts`.
 - Produces:
   - `PUBLICATION_NAME: string`, `PUBLICATION_DESCRIPTION: string`
@@ -2161,12 +2319,17 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 **`@atproto/api` usage, verified 2026-09-08 against https://www.npmjs.com/package/@atproto/api (latest published version **0.20.42**, read from `https://registry.npmjs.org/@atproto/api/latest`) and the package README at https://github.com/bluesky-social/atproto/tree/main/packages/api:**
 
 ```ts
-const agent = new AtpAgent({ service: "https://bsky.social" })
-await agent.login({ identifier, password })
-const res = await agent.com.atproto.repo.putRecord({ repo, collection, rkey, record })
+const agent = new AtpAgent({ service: "https://bsky.social" });
+await agent.login({ identifier, password });
+const res = await agent.com.atproto.repo.putRecord({
+  repo,
+  collection,
+  rkey,
+  record,
+});
 // res.data.uri, res.data.cid
-const who = await agent.com.atproto.identity.resolveHandle({ handle })  // who.data.did
-const rec = await agent.com.atproto.repo.getRecord({ repo, collection, rkey }) // rec.data.uri, rec.data.cid
+const who = await agent.com.atproto.identity.resolveHandle({ handle }); // who.data.did
+const rec = await agent.com.atproto.repo.getRecord({ repo, collection, rkey }); // rec.data.uri, rec.data.cid
 ```
 
 XRPC calls return `{ success, headers, data }`; the `uri`/`cid`/`did` live on `.data`. `agent.session.did` is the logged-in repo's DID after `login`.
@@ -2279,14 +2442,21 @@ test("outlineToText flattens the outline to indented 'Title — url' lines", ()
 
 test("bskyUrlToParts splits a post URL and rejects anything else", () => {
   assert.deepEqual(
-    bskyUrlToParts("https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x"),
+    bskyUrlToParts(
+      "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
+    ),
     { actor: "thismonthinreact.com", rkey: "3lqz7abcd2k2x" },
   );
   assert.deepEqual(
-    bskyUrlToParts("https://bsky.app/profile/did:plc:abc123/post/3lqz7abcd2k2x"),
+    bskyUrlToParts(
+      "https://bsky.app/profile/did:plc:abc123/post/3lqz7abcd2k2x",
+    ),
     { actor: "did:plc:abc123", rkey: "3lqz7abcd2k2x" },
   );
-  assert.equal(bskyUrlToParts("https://bsky.app/profile/thismonthinreact.com"), null);
+  assert.equal(
+    bskyUrlToParts("https://bsky.app/profile/thismonthinreact.com"),
+    null,
+  );
   assert.equal(bskyUrlToParts("https://example.com/whatever"), null);
 });
 
@@ -2358,7 +2528,8 @@ export function outlineToText(items: OutlineItem[], depth = 0): string {
   for (const item of items) {
     const indent = "  ".repeat(depth);
     lines.push(`${indent}${item.title}${item.url ? ` — ${item.url}` : ""}`);
-    if (item.children.length > 0) lines.push(outlineToText(item.children, depth + 1));
+    if (item.children.length > 0)
+      lines.push(outlineToText(item.children, depth + 1));
   }
   return lines.join("\n");
 }
@@ -2367,7 +2538,9 @@ export function outlineToText(items: OutlineItem[], depth = 0): string {
 export function bskyUrlToParts(
   url: string,
 ): { actor: string; rkey: string } | null {
-  const m = /^https:\/\/bsky\.app\/profile\/([^/]+)\/post\/([^/?#]+)/.exec(url.trim());
+  const m = /^https:\/\/bsky\.app\/profile\/([^/]+)\/post\/([^/?#]+)/.exec(
+    url.trim(),
+  );
   return m ? { actor: m[1], rkey: m[2] } : null;
 }
 
@@ -2427,7 +2600,8 @@ async function resolvePostRef(
   }
   const did = parts.actor.startsWith("did:")
     ? parts.actor
-    : (await agent.com.atproto.identity.resolveHandle({ handle: parts.actor })).data.did;
+    : (await agent.com.atproto.identity.resolveHandle({ handle: parts.actor }))
+        .data.did;
   const res = await agent.com.atproto.repo.getRecord({
     repo: did,
     collection: "app.bsky.feed.post",
@@ -2465,7 +2639,8 @@ async function main() {
     .filter((n) => n.endsWith(".md"))
     .filter((n) => !only || n === `${only}.md`)
     .sort();
-  if (only && names.length === 0) throw new Error(`no episode file for ${only}`);
+  if (only && names.length === 0)
+    throw new Error(`no episode file for ${only}`);
 
   for (const name of names) {
     const path = join(EPISODE_DIR, name);
@@ -2542,31 +2717,31 @@ Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>"
 
 **1. Spec coverage.**
 
-| Spec requirement | Task |
-| --- | --- |
-| `content/episodes/<yyyy>-<mm>.md` one file per episode | Task 3 (migrate writes them), format defined in "Canonical Episode File Format" |
-| Hand-written front matter: `title`, `date`, `description`, `time`, `location` | Task 2 (`Episode` fields, parser) |
-| Ingest-owned front matter: `transistorId`, `audioUrl`, `duration`, `season`, `episode`, `people`, `bskyPostUrl` | Task 4 (`applyFeedItem`) |
-| `atUri` written by `publish-atproto`, never by ingest | Task 7 (writer), Task 4 (asserted untouched) |
-| `site.standard.publication` at rkey `self`: `url`, `name`, `description` | Task 7 |
-| `site.standard.document` per episode keyed by slug: `site`, `title`, `publishedAt`, `path`, `description`, `textContent`, `tags`, `contributors`, `bskyPostRef` | Task 7 (`buildDocumentRecord`) |
-| `ATPROTO_HANDLE` / `ATPROTO_APP_PASSWORD`, `@atproto/api` | Task 7 (Global Constraints dependency note, `.env.example`) |
-| Ingest matches by `yyyy-mm` in the feed title, both title styles | Task 4 (`slugFromTitle`, tested against every real style) |
-| Body: outline then `# Transcript`, `##` sections, `**Speaker:**`, trailing `[hh:mm:ss]` | Task 2 |
-| Parsed structure `Episode`/`OutlineItem`/`Section`/`Segment` | Task 2 (interfaces block) |
-| `migrate` detects formats, strips embeds and style blocks, normalizes timestamps, promotes h1 sections to h2, synthesizes an outline, writes a report | Task 3 |
-| `ingest` idempotent, reports both directions of mismatch | Task 4 (idempotence test + Step 7, `no file for` / `no feed item for` lines) |
-| `publish-transcript` steps 1–4 with independent flags, `DESCRIPT_TOKEN` / `TRANSISTOR_API_KEY` | Task 6 |
-| SRT export with speaker prefixes | Task 5 |
-| Testing: parser fixtures per historical format; migrate round-trips through the parser; ingest against a saved feed; SRT cue numbering/format/prefix | Tasks 2–5 |
-| `/.well-known/site.standard.publication`, `<link rel="site.standard.document">`, the comments section | **Site plan**, `docs/superpowers/plans/2026-09-08-site.md` |
-| Site, routes, CSS, Pagefind, Netlify, newsletter, risk gate | **Out of scope**, per the plan header |
+| Spec requirement                                                                                                                                                | Task                                                                            |
+| --------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------- |
+| `content/episodes/<yyyy>-<mm>.md` one file per episode                                                                                                          | Task 3 (migrate writes them), format defined in "Canonical Episode File Format" |
+| Hand-written front matter: `title`, `date`, `description`, `time`, `location`                                                                                   | Task 2 (`Episode` fields, parser)                                               |
+| Ingest-owned front matter: `transistorId`, `audioUrl`, `duration`, `season`, `episode`, `people`, `bskyPostUrl`                                                 | Task 4 (`applyFeedItem`)                                                        |
+| `atUri` written by `publish-atproto`, never by ingest                                                                                                           | Task 7 (writer), Task 4 (asserted untouched)                                    |
+| `site.standard.publication` at rkey `self`: `url`, `name`, `description`                                                                                        | Task 7                                                                          |
+| `site.standard.document` per episode keyed by slug: `site`, `title`, `publishedAt`, `path`, `description`, `textContent`, `tags`, `contributors`, `bskyPostRef` | Task 7 (`buildDocumentRecord`)                                                  |
+| `ATPROTO_HANDLE` / `ATPROTO_APP_PASSWORD`, `@atproto/api`                                                                                                       | Task 7 (Global Constraints dependency note, `.env.example`)                     |
+| Ingest matches by `yyyy-mm` in the feed title, both title styles                                                                                                | Task 4 (`slugFromTitle`, tested against every real style)                       |
+| Body: outline then `# Transcript`, `##` sections, `**Speaker:**`, trailing `[hh:mm:ss]`                                                                         | Task 2                                                                          |
+| Parsed structure `Episode`/`OutlineItem`/`Section`/`Segment`                                                                                                    | Task 2 (interfaces block)                                                       |
+| `migrate` detects formats, strips embeds and style blocks, normalizes timestamps, promotes h1 sections to h2, synthesizes an outline, writes a report           | Task 3                                                                          |
+| `ingest` idempotent, reports both directions of mismatch                                                                                                        | Task 4 (idempotence test + Step 7, `no file for` / `no feed item for` lines)    |
+| `publish-transcript` steps 1–4 with independent flags, `DESCRIPT_TOKEN` / `TRANSISTOR_API_KEY`                                                                  | Task 6                                                                          |
+| SRT export with speaker prefixes                                                                                                                                | Task 5                                                                          |
+| Testing: parser fixtures per historical format; migrate round-trips through the parser; ingest against a saved feed; SRT cue numbering/format/prefix            | Tasks 2–5                                                                       |
+| `/.well-known/site.standard.publication`, `<link rel="site.standard.document">`, the comments section                                                           | **Site plan**, `docs/superpowers/plans/2026-09-08-site.md`                      |
+| Site, routes, CSS, Pagefind, Netlify, newsletter, risk gate                                                                                                     | **Out of scope**, per the plan header                                           |
 
 Two gaps found and closed while reviewing: `TRANSISTOR_SHOW_ID` is needed to resolve a share id to an episode id and is not in the spec's env list — added to Task 6's environment note. And the spec's "three historical formats" is not accurate to the corpus; Task 3 replaces it with a per-feature detection table derived from reading all 39 files, and calls out the two files (`tmir-2024-04.md`, `tmir-2025-06.md`) that break the naive rules.
 
 **2. Placeholder scan.** No TBD/TODO, no "add error handling", no "similar to Task N". Every code step carries runnable code. The one deliberate uncertainty is the Descript markdown export shape, which is labelled as such, given a concrete assumed shape from a real pasted export, isolated to a single function, and paired with a verification step (Task 6, Step 6). Task 7 carries a second, equally explicit one: the document lexicon marks `contributors[].did` required and the feed gives us no DIDs, so the record is written without it and Step 9 names both fallbacks to take if the PDS rejects it.
 
-**3. Type consistency.** Checked across tasks: `slug`/`flattenLinks`/`normalizeTime`/`secondsToTimestamp`/`timestampToSeconds` (Task 1) are used with the same names in Tasks 2, 3, 5, 6. `splitFile` returns `{ frontMatter, body }` in Task 2 and is destructured that way in Tasks 3 and 4. `serializeEpisodeFile(frontMatter, body)` has the same two-argument signature everywhere. `Person` is defined in `parse.ts` and imported by `ingest.ts` as a type-only import. `TRANSCRIPT_MARKER` is exported from `parse.ts` and consumed in Task 6. `toSrt(episode)` takes the `Episode` from `parseEpisode`. 
+**3. Type consistency.** Checked across tasks: `slug`/`flattenLinks`/`normalizeTime`/`secondsToTimestamp`/`timestampToSeconds` (Task 1) are used with the same names in Tasks 2, 3, 5, 6. `splitFile` returns `{ frontMatter, body }` in Task 2 and is destructured that way in Tasks 3 and 4. `serializeEpisodeFile(frontMatter, body)` has the same two-argument signature everywhere. `Person` is defined in `parse.ts` and imported by `ingest.ts` as a type-only import. `TRANSCRIPT_MARKER` is exported from `parse.ts` and consumed in Task 6. `toSrt(episode)` takes the `Episode` from `parseEpisode`.
 Three problems found and fixed inline while reviewing:
 
 1. Task 6 imported `splitFile` without using it — removed from the import list.
diff --git a/docs/superpowers/plans/2026-09-08-site.md b/docs/superpowers/plans/2026-09-08-site.md
index b2ca018..65fb3aa 100644
--- a/docs/superpowers/plans/2026-09-08-site.md
+++ b/docs/superpowers/plans/2026-09-08-site.md
@@ -28,8 +28,8 @@ These apply to every task. Do not restate them; do not violate them.
 - **Do not deploy and do not push.** Stop at the build artifact. `netlify.toml` is written for Netlify to consume later; running `netlify deploy` is out of scope.
 - **Transcripts must never be serialized into the HTML twice.** RSC payloads and router loader data are both inlined for hydration, which doubles an hour-long transcript page from ~100 KB to ~200 KB. That is why `/episodes/<slug>` and `/links` are static documents. Do not "simplify" them into router routes with loaders.
 - **A route file cannot have both a GET handler returning HTML and a `component` on the same path.** Static-document routes have `server.handlers.GET` and nothing else.
-- **Static-document pages are outside the router.** Every link *to* one — from the nav, the home archive, anywhere — must be a plain `<a href="/episodes/…">`, never `<Link>`. The nav is plain `<a href>` throughout, because half its targets are outside the router.
-- **Server-only code must be dynamically imported inside handlers.** Route module files are part of the client route tree. Anything that touches `node:fs` — `src/content/load.ts`, `src/content/links.ts`, `src/content/feed.ts` — must be reached with `await import('...')` *inside* a handler, never with a top-level `import` in a `src/routes/*` file. Pure modules (`src/content/time.ts`, `src/content/types.ts`, `src/components/*`) may be imported normally.
+- **Static-document pages are outside the router.** Every link _to_ one — from the nav, the home archive, anywhere — must be a plain `<a href="/episodes/…">`, never `<Link>`. The nav is plain `<a href>` throughout, because half its targets are outside the router.
+- **Server-only code must be dynamically imported inside handlers.** Route module files are part of the client route tree. Anything that touches `node:fs` — `src/content/load.ts`, `src/content/links.ts`, `src/content/feed.ts` — must be reached with `await import('...')` _inside_ a handler, never with a top-level `import` in a `src/routes/*` file. Pure modules (`src/content/time.ts`, `src/content/types.ts`, `src/components/*`) may be imported normally.
 - **The stylesheet is a static file at `public/styles.css`**, referenced as `/styles.css`. It is not imported through Vite, because the static-document path has no Vite module graph to hash a URL through and both paths must reference the same URL.
 - **Dynamic routes and non-HTML routes need explicit `pages[]` entries** in `vite.config.ts`. The prerender crawler only follows `<a href="/...">` — not `<link>`, `<script>`, or fetches.
 - **`time` values are seconds as `number | undefined`** throughout the parsed `Episode` structure. Task 2 verifies this against the real `src/content/parse.ts`; if the pipeline emits `"hh:mm:ss"` strings instead, `toSeconds` in `src/content/time.ts` absorbs the difference and nothing else changes.
@@ -46,55 +46,57 @@ These apply to every task. Do not restate them; do not violate them.
 
 ## File Structure
 
-| File | Responsibility |
-| --- | --- |
-| `package.json` | merged: pipeline scripts + site deps and scripts |
-| `vite.config.ts` | plugin order, RSC, prerender page list generated from `content/episodes/` |
-| `tsconfig.json`, `tsr.config.json` | TS + router codegen config |
-| `netlify.toml` | build command (vite build then pagefind), publish dir |
-| `public/styles.css` | the entire stylesheet, served statically |
-| `scripts/check-build.mjs` | build assertions and size measurement; grows one block per task |
-| `src/content/time.ts` | `hms`, `toSeconds` — pure, safe to import from components |
-| `src/content/atproto.ts` | `bskyPostToAtUri` — pure, build-time `bsky.app` URL → `at://` URI |
+| File                                           | Responsibility                                                                     |
+| ---------------------------------------------- | ---------------------------------------------------------------------------------- |
+| `package.json`                                 | merged: pipeline scripts + site deps and scripts                                   |
+| `vite.config.ts`                               | plugin order, RSC, prerender page list generated from `content/episodes/`          |
+| `tsconfig.json`, `tsr.config.json`             | TS + router codegen config                                                         |
+| `netlify.toml`                                 | build command (vite build then pagefind), publish dir                              |
+| `public/styles.css`                            | the entire stylesheet, served statically                                           |
+| `scripts/check-build.mjs`                      | build assertions and size measurement; grows one block per task                    |
+| `src/content/time.ts`                          | `hms`, `toSeconds` — pure, safe to import from components                          |
+| `src/content/atproto.ts`                       | `bskyPostToAtUri` — pure, build-time `bsky.app` URL → `at://` URI                  |
 | `public/.well-known/site.standard.publication` | written at build time from `VITE_ATPROTO_PUBLICATION_URI`, copied to `dist/client` |
-| `src/content/load.ts` | `loadEpisodes`, `getEpisode` — reads the filesystem |
-| `src/content/links.ts` | `buildLinkIndex`, `groupByHost` |
-| `src/content/feed.ts` | `outlineToHtml`, `outlineToChapters`, `renderFeed` |
-| `src/components/Document.tsx` | shared `<html>` shell: head, stylesheet link, nav, `<main>` |
-| `src/components/EpisodeBody.tsx` | `Outline` and `EpisodeBody` markup, used by both paths |
-| `src/components/Player.tsx` | `'use client'` — home page player only |
-| `src/components/PagefindUI.tsx` | `'use client'` — `/search` only |
-| `src/routes/__root.tsx` | router shell, wraps `Document` with head/scripts slots |
-| `src/routes/index.tsx` | `/` — router page |
-| `src/routes/episodes.$slug.tsx` | `/episodes/<slug>` — static document |
-| `src/routes/links.tsx` | `/links` — static document |
-| `src/routes/search.tsx` | `/search` — router page |
-| `src/routes/about.tsx` | `/about` — router page |
-| `src/routes/feed[.]xml.ts` | `/feed.xml` — server route |
-| `src/routes/episodes.$slug.chapters[.]json.ts` | chapters JSON — server route |
+| `src/content/load.ts`                          | `loadEpisodes`, `getEpisode` — reads the filesystem                                |
+| `src/content/links.ts`                         | `buildLinkIndex`, `groupByHost`                                                    |
+| `src/content/feed.ts`                          | `outlineToHtml`, `outlineToChapters`, `renderFeed`                                 |
+| `src/components/Document.tsx`                  | shared `<html>` shell: head, stylesheet link, nav, `<main>`                        |
+| `src/components/EpisodeBody.tsx`               | `Outline` and `EpisodeBody` markup, used by both paths                             |
+| `src/components/Player.tsx`                    | `'use client'` — home page player only                                             |
+| `src/components/PagefindUI.tsx`                | `'use client'` — `/search` only                                                    |
+| `src/routes/__root.tsx`                        | router shell, wraps `Document` with head/scripts slots                             |
+| `src/routes/index.tsx`                         | `/` — router page                                                                  |
+| `src/routes/episodes.$slug.tsx`                | `/episodes/<slug>` — static document                                               |
+| `src/routes/links.tsx`                         | `/links` — static document                                                         |
+| `src/routes/search.tsx`                        | `/search` — router page                                                            |
+| `src/routes/about.tsx`                         | `/about` — router page                                                             |
+| `src/routes/feed[.]xml.ts`                     | `/feed.xml` — server route                                                         |
+| `src/routes/episodes.$slug.chapters[.]json.ts` | chapters JSON — server route                                                       |
 
 ---
 
 ## Task 1: Scaffold, build config, shared Document, root route
 
 **Files:**
+
 - Create or Modify: `package.json`
 - Create: `vite.config.ts`, `tsconfig.json`, `tsr.config.json`, `netlify.toml`, `.gitignore` (modify if present), `public/styles.css`, `src/components/Document.tsx`, `src/routes/__root.tsx`, `src/routes/index.tsx`, `scripts/check-build.mjs`
 - Test: `scripts/check-build.mjs`
 
 **Interfaces:**
+
 - Consumes: nothing.
 - Produces:
   - `SITE_NAME: string`, `SITE_URL: string` and `<Document>` from `src/components/Document.tsx`:
     ```ts
     function Document(props: {
-      head: React.ReactNode       // router path passes <HeadContent/>; static path passes <title>/<meta>/<link rel=canonical>
-      scripts?: React.ReactNode   // router path passes <Scripts/>; static path passes its inline script
-      bodyClass?: string
-      children: React.ReactNode
-    }): JSX.Element
+      head: React.ReactNode; // router path passes <HeadContent/>; static path passes <title>/<meta>/<link rel=canonical>
+      scripts?: React.ReactNode; // router path passes <Scripts/>; static path passes its inline script
+      bodyClass?: string;
+      children: React.ReactNode;
+    }): JSX.Element;
     ```
-    **`Document` deliberately renders no `<title>` of its own.** Verified in the spike: when the shell emits a `<title>` *and* `<HeadContent/>` emits the route's, both tags land in the HTML and browsers use the **first** one — so a shell-owned title would silently override every route title. The `head` slot is the single source of per-page head tags.
+    **`Document` deliberately renders no `<title>` of its own.** Verified in the spike: when the shell emits a `<title>` _and_ `<HeadContent/>` emits the route's, both tags land in the HTML and browsers use the **first** one — so a shell-owned title would silently override every route title. The `head` slot is the single source of per-page head tags.
   - A buildable app. `npm run build` emits `dist/client/`. `node scripts/check-build.mjs` asserts the output and prints sizes.
   - `src/routes/index.tsx` is a stub replaced in Task 4; `public/styles.css` is a stub replaced in Task 9.
 
@@ -105,69 +107,71 @@ These apply to every task. Do not restate them; do not violate them.
 Create `scripts/check-build.mjs`:
 
 ```js
-import { readFileSync, existsSync, readdirSync, statSync } from 'node:fs'
-import { gzipSync } from 'node:zlib'
-import { join } from 'node:path'
+import { readFileSync, existsSync, readdirSync, statSync } from "node:fs";
+import { gzipSync } from "node:zlib";
+import { join } from "node:path";
 
-const dist = 'dist/client'
-let failures = 0
+const dist = "dist/client";
+let failures = 0;
 
-function check(label, ok, detail = '') {
+function check(label, ok, detail = "") {
   if (ok) {
-    console.log(`  ok   ${label}`)
+    console.log(`  ok   ${label}`);
   } else {
-    failures++
-    console.log(`  FAIL ${label}${detail ? ` — ${detail}` : ''}`)
+    failures++;
+    console.log(`  FAIL ${label}${detail ? ` — ${detail}` : ""}`);
   }
 }
 
 function file(path) {
-  const full = join(dist, path)
-  return existsSync(full) ? readFileSync(full, 'utf8') : null
+  const full = join(dist, path);
+  return existsSync(full) ? readFileSync(full, "utf8") : null;
 }
 
 function hasFile(path) {
-  check(`${path} exists`, file(path) !== null)
+  check(`${path} exists`, file(path) !== null);
 }
 
 function contains(path, needle) {
-  const body = file(path)
+  const body = file(path);
   check(
     `${path} contains ${JSON.stringify(needle)}`,
     body !== null && body.includes(needle),
-    body === null ? 'file missing' : 'needle absent',
-  )
+    body === null ? "file missing" : "needle absent",
+  );
 }
 
 function slugs() {
-  return readdirSync('content/episodes')
-    .filter((f) => f.endsWith('.md'))
-    .map((f) => f.slice(0, -'.md'.length))
-    .sort()
+  return readdirSync("content/episodes")
+    .filter((f) => f.endsWith(".md"))
+    .map((f) => f.slice(0, -".md".length))
+    .sort();
 }
 
-console.log('Task 1: shell')
-hasFile('index.html')
-hasFile('styles.css')
-contains('index.html', 'class="site-nav"')
-contains('index.html', 'href="/styles.css"')
+console.log("Task 1: shell");
+hasFile("index.html");
+hasFile("styles.css");
+contains("index.html", 'class="site-nav"');
+contains("index.html", 'href="/styles.css"');
 
-console.log('\nSizes')
-const assets = join(dist, 'assets')
+console.log("\nSizes");
+const assets = join(dist, "assets");
 const entry = existsSync(assets)
   ? readdirSync(assets)
-      .filter((f) => f.endsWith('.js'))
+      .filter((f) => f.endsWith(".js"))
       .map((f) => ({ f, size: statSync(join(assets, f)).size }))
       .sort((a, b) => b.size - a.size)[0]
-  : null
+  : null;
 if (entry) {
-  const raw = readFileSync(join(assets, entry.f))
-  console.log(`  router entry ${entry.f}: ${raw.length} B raw / ${gzipSync(raw).length} B gzip`)
+  const raw = readFileSync(join(assets, entry.f));
+  console.log(
+    `  router entry ${entry.f}: ${raw.length} B raw / ${gzipSync(raw).length} B gzip`,
+  );
 } else {
-  console.log('  router entry: none found')
+  console.log("  router entry: none found");
 }
 
-process.exit(failures === 0 ? 0 : 1)
+process.exit(failures === 0 ? 0 : 1);
 ```
 
 - [ ] **Step 2: Run it to verify it fails**
@@ -256,42 +260,42 @@ The `pages` array is generated by reading the content directory at config load,
 A five-line inline plugin writes `/.well-known/site.standard.publication` into `public/` at `buildStart`, so Vite's own public-directory copy lands it in `dist/client/.well-known/`. That is the whole mechanism — no post-build script, no extra prerender entry. The file is generated, so `.gitignore` gets `public/.well-known/` in Step 7.
 
 ```ts
-import { readdirSync, existsSync, mkdirSync, writeFileSync } from 'node:fs'
-import { defineConfig } from 'vite'
-import { tanstackStart } from '@tanstack/react-start/plugin/vite'
-import viteReact from '@vitejs/plugin-react'
-import rsc from '@vitejs/plugin-rsc'
-import netlify from '@netlify/vite-plugin-tanstack-start'
-
-const episodesDir = 'content/episodes'
+import { readdirSync, existsSync, mkdirSync, writeFileSync } from "node:fs";
+import { defineConfig } from "vite";
+import { tanstackStart } from "@tanstack/react-start/plugin/vite";
+import viteReact from "@vitejs/plugin-react";
+import rsc from "@vitejs/plugin-rsc";
+import netlify from "@netlify/vite-plugin-tanstack-start";
+
+const episodesDir = "content/episodes";
 const slugs = existsSync(episodesDir)
   ? readdirSync(episodesDir)
-      .filter((f) => f.endsWith('.md'))
-      .map((f) => f.slice(0, -'.md'.length))
+      .filter((f) => f.endsWith(".md"))
+      .map((f) => f.slice(0, -".md".length))
       .sort()
-  : []
+  : [];
 
 const pages = [
-  { path: '/feed.xml', prerender: { enabled: true } },
+  { path: "/feed.xml", prerender: { enabled: true } },
   ...slugs.flatMap((slug) => [
     { path: `/episodes/${slug}`, prerender: { enabled: true } },
     { path: `/episodes/${slug}/chapters.json`, prerender: { enabled: true } },
   ]),
-]
+];
 
 /**
  * standard.site discovery file. Written into public/ so Vite's normal public
  * copy puts it at dist/client/.well-known/site.standard.publication.
  */
 const wellKnownPublication = {
-  name: 'well-known-publication',
+  name: "well-known-publication",
   buildStart() {
-    const uri = process.env.VITE_ATPROTO_PUBLICATION_URI
-    if (!uri) return
-    mkdirSync('public/.well-known', { recursive: true })
-    writeFileSync('public/.well-known/site.standard.publication', `${uri}\n`)
+    const uri = process.env.VITE_ATPROTO_PUBLICATION_URI;
+    if (!uri) return;
+    mkdirSync("public/.well-known", { recursive: true });
+    writeFileSync("public/.well-known/site.standard.publication", `${uri}\n`);
   },
-}
+};
 
 export default defineConfig({
   resolve: { tsconfigPaths: true },
@@ -308,7 +312,7 @@ export default defineConfig({
     rsc(),
     viteReact(),
   ],
-})
+});
 ```
 
 - [ ] **Step 6: Write `netlify.toml`**
@@ -345,10 +349,11 @@ public/.well-known/
 This is a plain component with no server-only imports, so both the router shell and the static handlers can use it. The nav uses plain `<a href>` because `/episodes/…` and `/links` are outside the router.
 
 ```tsx
-export const SITE_NAME = import.meta.env.VITE_SITE_NAME || 'This Month in React'
+export const SITE_NAME =
+  import.meta.env.VITE_SITE_NAME || "This Month in React";
 export const SITE_URL = (
-  import.meta.env.VITE_SITE_URL || 'https://thismonthinreact.com'
-).replace(/\/$/, '')
+  import.meta.env.VITE_SITE_URL || "https://thismonthinreact.com"
+).replace(/\/$/, "");
 
 /**
  * Renders no <title> itself. The router shell passes <HeadContent/> here and the
@@ -362,10 +367,10 @@ export function Document({
   bodyClass,
   children,
 }: {
-  head: React.ReactNode
-  scripts?: React.ReactNode
-  bodyClass?: string
-  children: React.ReactNode
+  head: React.ReactNode;
+  scripts?: React.ReactNode;
+  bodyClass?: string;
+  children: React.ReactNode;
 }) {
   return (
     <html lang="en">
@@ -394,14 +399,16 @@ export function Document({
             <a href="/about">About</a>
           </nav>
         </header>
-        <main className={bodyClass ? `${bodyClass}-main` : undefined}>{children}</main>
+        <main className={bodyClass ? `${bodyClass}-main` : undefined}>
+          {children}
+        </main>
         <footer className="site-footer" data-pagefind-ignore="">
           <a href="/feed.xml">RSS</a>
         </footer>
         {scripts}
       </body>
     </html>
-  )
+  );
 }
 ```
 
@@ -412,20 +419,20 @@ export function Document({
 The router's shell renders the same `Document`, injecting `<HeadContent/>` and `<Scripts/>` through the slots.
 
 ```tsx
-import { HeadContent, Scripts, createRootRoute } from '@tanstack/react-router'
-import { Document, SITE_NAME } from '../components/Document'
+import { HeadContent, Scripts, createRootRoute } from "@tanstack/react-router";
+import { Document, SITE_NAME } from "../components/Document";
 
 export const Route = createRootRoute({
   head: () => ({ meta: [{ title: SITE_NAME }] }),
   shellComponent: RootDocument,
-})
+});
 
 function RootDocument({ children }: { children: React.ReactNode }) {
   return (
     <Document head={<HeadContent />} scripts={<Scripts />}>
       {children}
     </Document>
-  )
+  );
 }
 ```
 
@@ -436,12 +443,12 @@ function RootDocument({ children }: { children: React.ReactNode }) {
 `src/routes/index.tsx` (replaced wholesale in Task 4):
 
 ```tsx
-import { createFileRoute } from '@tanstack/react-router'
+import { createFileRoute } from "@tanstack/react-router";
 
-export const Route = createFileRoute('/')({ component: Home })
+export const Route = createFileRoute("/")({ component: Home });
 
 function Home() {
-  return <h1>Episodes</h1>
+  return <h1>Episodes</h1>;
 }
 ```
 
@@ -480,27 +487,58 @@ EOF
 ## Task 2: Time helpers, episode loading, link index
 
 **Files:**
+
 - Create: `src/content/time.ts`, `src/content/load.ts`, `src/content/links.ts`
 - Test: `src/content/time.test.ts`, `src/content/load.test.ts`, `src/content/links.test.ts`
 - Test fixture: `test/fixtures/episodes/2026-05.md`
 
 **Interfaces:**
+
 - Consumes, from the content pipeline plan:
   - `src/content/parse.ts` exporting `parseEpisode(markdown: string): Episode`
   - `src/content/types.ts` exporting:
     ```ts
-    interface Person { name: string; role?: string; href?: string; img?: string }
-    interface OutlineItem { title: string; url?: string; time?: number; anchor: string; children: OutlineItem[] }
-    interface Segment { speaker?: string; time?: number; text: string }
-    interface Section { title: string; anchor: string; time?: number; segments: Segment[] }
+    interface Person {
+      name: string;
+      role?: string;
+      href?: string;
+      img?: string;
+    }
+    interface OutlineItem {
+      title: string;
+      url?: string;
+      time?: number;
+      anchor: string;
+      children: OutlineItem[];
+    }
+    interface Segment {
+      speaker?: string;
+      time?: number;
+      text: string;
+    }
+    interface Section {
+      title: string;
+      anchor: string;
+      time?: number;
+      segments: Segment[];
+    }
     interface Episode {
-      slug: string; title: string; date: string; description: string
-      time?: string; location?: string
-      transistorId?: string; audioUrl?: string; duration?: number
-      season?: number; episode?: number
-      bskyPostUrl?: string   // announcement post, written by ingest
-      atUri?: string         // site.standard.document URI, written by publish-atproto
-      people: Person[]; outline: OutlineItem[]; sections: Section[]
+      slug: string;
+      title: string;
+      date: string;
+      description: string;
+      time?: string;
+      location?: string;
+      transistorId?: string;
+      audioUrl?: string;
+      duration?: number;
+      season?: number;
+      episode?: number;
+      bskyPostUrl?: string; // announcement post, written by ingest
+      atUri?: string; // site.standard.document URI, written by publish-atproto
+      people: Person[];
+      outline: OutlineItem[];
+      sections: Section[];
     }
     ```
 - Produces:
@@ -572,95 +610,95 @@ It took a while. [00:00:45]
 `src/content/time.test.ts`:
 
 ```ts
-import { test } from 'node:test'
-import assert from 'node:assert/strict'
-import { hms, toSeconds } from './time.ts'
-
-test('hms formats seconds', () => {
-  assert.equal(hms(30), '0:30')
-  assert.equal(hms(123), '2:03')
-  assert.equal(hms(3723), '1:02:03')
-})
-
-test('toSeconds accepts numbers and both string shapes', () => {
-  assert.equal(toSeconds(90), 90)
-  assert.equal(toSeconds('00:01:30'), 90)
-  assert.equal(toSeconds('01:30'), 90)
-  assert.equal(toSeconds(undefined), undefined)
-  assert.equal(toSeconds('nonsense'), undefined)
-})
+import { test } from "node:test";
+import assert from "node:assert/strict";
+import { hms, toSeconds } from "./time.ts";
+
+test("hms formats seconds", () => {
+  assert.equal(hms(30), "0:30");
+  assert.equal(hms(123), "2:03");
+  assert.equal(hms(3723), "1:02:03");
+});
+
+test("toSeconds accepts numbers and both string shapes", () => {
+  assert.equal(toSeconds(90), 90);
+  assert.equal(toSeconds("00:01:30"), 90);
+  assert.equal(toSeconds("01:30"), 90);
+  assert.equal(toSeconds(undefined), undefined);
+  assert.equal(toSeconds("nonsense"), undefined);
+});
 ```
 
 `src/content/load.test.ts`:
 
 ```ts
-import { test } from 'node:test'
-import assert from 'node:assert/strict'
-
-process.env.TMIR_CONTENT_DIR = 'test/fixtures/episodes'
-const { loadEpisodes, getEpisode } = await import('./load.ts')
-
-test('loadEpisodes reads and parses the content directory', async () => {
-  const episodes = await loadEpisodes()
-  assert.equal(episodes.length, 1)
-  assert.equal(episodes[0].slug, '2026-05')
-  assert.equal(episodes[0].title, 'TMiR 2026-05: React Compiler ships')
-  assert.equal(episodes[0].sections.length, 2)
-})
-
-test('loadEpisodes sorts by date descending', async () => {
-  const dates = (await loadEpisodes()).map((e) => e.date)
-  assert.deepEqual(dates, [...dates].sort().reverse())
-})
-
-test('getEpisode finds by slug and misses cleanly', async () => {
-  assert.equal((await getEpisode('2026-05'))?.slug, '2026-05')
-  assert.equal(await getEpisode('1999-01'), undefined)
-})
+import { test } from "node:test";
+import assert from "node:assert/strict";
+
+process.env.TMIR_CONTENT_DIR = "test/fixtures/episodes";
+const { loadEpisodes, getEpisode } = await import("./load.ts");
+
+test("loadEpisodes reads and parses the content directory", async () => {
+  const episodes = await loadEpisodes();
+  assert.equal(episodes.length, 1);
+  assert.equal(episodes[0].slug, "2026-05");
+  assert.equal(episodes[0].title, "TMiR 2026-05: React Compiler ships");
+  assert.equal(episodes[0].sections.length, 2);
+});
+
+test("loadEpisodes sorts by date descending", async () => {
+  const dates = (await loadEpisodes()).map((e) => e.date);
+  assert.deepEqual(dates, [...dates].sort().reverse());
+});
+
+test("getEpisode finds by slug and misses cleanly", async () => {
+  assert.equal((await getEpisode("2026-05"))?.slug, "2026-05");
+  assert.equal(await getEpisode("1999-01"), undefined);
+});
 ```
 
 `src/content/links.test.ts`:
 
 ```ts
-import { test } from 'node:test'
-import assert from 'node:assert/strict'
+import { test } from "node:test";
+import assert from "node:assert/strict";
 
-process.env.TMIR_CONTENT_DIR = 'test/fixtures/episodes'
-const { buildLinkIndex, groupByHost } = await import('./links.ts')
-const { loadEpisodes } = await import('./load.ts')
+process.env.TMIR_CONTENT_DIR = "test/fixtures/episodes";
+const { buildLinkIndex, groupByHost } = await import("./links.ts");
+const { loadEpisodes } = await import("./load.ts");
 
-test('buildLinkIndex flattens nested outline items with URLs', async () => {
-  const entries = buildLinkIndex(await loadEpisodes())
+test("buildLinkIndex flattens nested outline items with URLs", async () => {
+  const entries = buildLinkIndex(await loadEpisodes());
   assert.deepEqual(
     entries.map((e) => e.text),
-    ['React Compiler', 'Caveats', 'TanStack Start'],
-  )
-})
-
-test('buildLinkIndex excludes items without a URL', async () => {
-  const entries = buildLinkIndex(await loadEpisodes())
-  assert.ok(!entries.some((e) => e.text === 'No link here'))
-})
-
-test('buildLinkIndex records host, episode and time', async () => {
-  const [first] = buildLinkIndex(await loadEpisodes())
-  assert.equal(first.host, 'react.dev')
-  assert.equal(first.url, 'https://react.dev/compiler')
-  assert.equal(first.episodeSlug, '2026-05')
-  assert.equal(first.episodeTitle, 'TMiR 2026-05: React Compiler ships')
-  assert.equal(first.time, 30)
-})
-
-test('groupByHost sorts by count descending', async () => {
-  const groups = groupByHost(buildLinkIndex(await loadEpisodes()))
+    ["React Compiler", "Caveats", "TanStack Start"],
+  );
+});
+
+test("buildLinkIndex excludes items without a URL", async () => {
+  const entries = buildLinkIndex(await loadEpisodes());
+  assert.ok(!entries.some((e) => e.text === "No link here"));
+});
+
+test("buildLinkIndex records host, episode and time", async () => {
+  const [first] = buildLinkIndex(await loadEpisodes());
+  assert.equal(first.host, "react.dev");
+  assert.equal(first.url, "https://react.dev/compiler");
+  assert.equal(first.episodeSlug, "2026-05");
+  assert.equal(first.episodeTitle, "TMiR 2026-05: React Compiler ships");
+  assert.equal(first.time, 30);
+});
+
+test("groupByHost sorts by count descending", async () => {
+  const groups = groupByHost(buildLinkIndex(await loadEpisodes()));
   assert.deepEqual(
     groups.map((g) => [g.host, g.entries.length]),
     [
-      ['react.dev', 2],
-      ['tanstack.com', 1],
+      ["react.dev", 2],
+      ["tanstack.com", 1],
     ],
-  )
-})
+  );
+});
 ```
 
 > The dynamic `await import` in the two filesystem tests matters: `TMIR_CONTENT_DIR` must be set before `load.ts` first resolves the directory.
@@ -673,20 +711,24 @@ Expected: FAIL — `Cannot find module './time.ts'`.
 - [ ] **Step 5: Write `src/content/time.ts`**
 
 ```ts
-export function toSeconds(value: string | number | undefined): number | undefined {
-  if (value === undefined) return undefined
-  if (typeof value === 'number') return Number.isFinite(value) ? value : undefined
-  const parts = value.split(':').map(Number)
-  if (parts.length === 0 || parts.some((n) => !Number.isFinite(n))) return undefined
-  return parts.reduce((total, part) => total * 60 + part, 0)
+export function toSeconds(
+  value: string | number | undefined,
+): number | undefined {
+  if (value === undefined) return undefined;
+  if (typeof value === "number")
+    return Number.isFinite(value) ? value : undefined;
+  const parts = value.split(":").map(Number);
+  if (parts.length === 0 || parts.some((n) => !Number.isFinite(n)))
+    return undefined;
+  return parts.reduce((total, part) => total * 60 + part, 0);
 }
 
 export function hms(seconds: number): string {
-  const s = Math.max(0, Math.floor(seconds))
-  const h = Math.floor(s / 3600)
-  const m = Math.floor((s % 3600) / 60)
-  const sec = String(s % 60).padStart(2, '0')
-  return h > 0 ? `${h}:${String(m).padStart(2, '0')}:${sec}` : `${m}:${sec}`
+  const s = Math.max(0, Math.floor(seconds));
+  const h = Math.floor(s / 3600);
+  const m = Math.floor((s % 3600) / 60);
+  const sec = String(s % 60).padStart(2, "0");
+  return h > 0 ? `${h}:${String(m).padStart(2, "0")}:${sec}` : `${m}:${sec}`;
 }
 ```
 
@@ -695,60 +737,65 @@ export function hms(seconds: number): string {
 `TMIR_CONTENT_DIR` exists only so tests can point at fixtures; production always uses the default. `process.cwd()` is correct at prerender time (verified by the spike).
 
 ```ts
-import { readdir, readFile } from 'node:fs/promises'
-import { join } from 'node:path'
-import { parseEpisode } from './parse.ts'
-import type { Episode } from './types.ts'
+import { readdir, readFile } from "node:fs/promises";
+import { join } from "node:path";
+import { parseEpisode } from "./parse.ts";
+import type { Episode } from "./types.ts";
 
 function contentDir(): string {
-  return join(process.cwd(), process.env.TMIR_CONTENT_DIR || 'content/episodes')
+  return join(
+    process.cwd(),
+    process.env.TMIR_CONTENT_DIR || "content/episodes",
+  );
 }
 
-let cache: Promise<Episode[]> | undefined
+let cache: Promise<Episode[]> | undefined;
 
 export function loadEpisodes(): Promise<Episode[]> {
   cache ??= (async () => {
-    const dir = contentDir()
-    const files = (await readdir(dir)).filter((f) => f.endsWith('.md')).sort()
+    const dir = contentDir();
+    const files = (await readdir(dir)).filter((f) => f.endsWith(".md")).sort();
     const episodes = await Promise.all(
       files.map(async (f) => {
-        const markdown = await readFile(join(dir, f), 'utf8')
-        return { ...parseEpisode(markdown), slug: f.slice(0, -'.md'.length) }
+        const markdown = await readFile(join(dir, f), "utf8");
+        return { ...parseEpisode(markdown), slug: f.slice(0, -".md".length) };
       }),
-    )
-    return episodes.sort((a, b) => (a.date < b.date ? 1 : a.date > b.date ? -1 : 0))
-  })()
-  return cache
+    );
+    return episodes.sort((a, b) =>
+      a.date < b.date ? 1 : a.date > b.date ? -1 : 0,
+    );
+  })();
+  return cache;
 }
 
 export async function getEpisode(slug: string): Promise<Episode | undefined> {
-  return (await loadEpisodes()).find((e) => e.slug === slug)
+  return (await loadEpisodes()).find((e) => e.slug === slug);
 }
 ```
 
 - [ ] **Step 7: Write `src/content/links.ts`**
 
 ```ts
-import type { Episode, OutlineItem } from './types.ts'
-import { toSeconds } from './time.ts'
+import type { Episode, OutlineItem } from "./types.ts";
+import { toSeconds } from "./time.ts";
 
 export type LinkEntry = {
-  text: string
-  url: string
-  host: string
-  episodeSlug: string
-  episodeTitle: string
-  time?: number
-}
+  text: string;
+  url: string;
+  host: string;
+  episodeSlug: string;
+  episodeTitle: string;
+  time?: number;
+};
 
 function walk(items: OutlineItem[], episode: Episode, out: LinkEntry[]): void {
   for (const item of items) {
     if (item.url) {
-      let host: string
+      let host: string;
       try {
-        host = new URL(item.url).hostname.replace(/^www\./, '')
+        host = new URL(item.url).hostname.replace(/^www\./, "");
       } catch {
-        host = 'unknown'
+        host = "unknown";
       }
       out.push({
         text: item.title,
@@ -757,28 +804,33 @@ function walk(items: OutlineItem[], episode: Episode, out: LinkEntry[]): void {
         episodeSlug: episode.slug,
         episodeTitle: episode.title,
         time: toSeconds(item.time),
-      })
+      });
     }
-    walk(item.children, episode, out)
+    walk(item.children, episode, out);
   }
 }
 
 export function buildLinkIndex(episodes: Episode[]): LinkEntry[] {
-  const out: LinkEntry[] = []
-  for (const episode of episodes) walk(episode.outline, episode, out)
-  return out
+  const out: LinkEntry[] = [];
+  for (const episode of episodes) walk(episode.outline, episode, out);
+  return out;
 }
 
-export function groupByHost(entries: LinkEntry[]): { host: string; entries: LinkEntry[] }[] {
-  const map = new Map<string, LinkEntry[]>()
+export function groupByHost(
+  entries: LinkEntry[],
+): { host: string; entries: LinkEntry[] }[] {
+  const map = new Map<string, LinkEntry[]>();
   for (const entry of entries) {
-    const list = map.get(entry.host)
-    if (list) list.push(entry)
-    else map.set(entry.host, [entry])
+    const list = map.get(entry.host);
+    if (list) list.push(entry);
+    else map.set(entry.host, [entry]);
   }
   return [...map]
     .map(([host, list]) => ({ host, entries: list }))
-    .sort((a, b) => b.entries.length - a.entries.length || a.host.localeCompare(b.host))
+    .sort(
+      (a, b) =>
+        b.entries.length - a.entries.length || a.host.localeCompare(b.host),
+    );
 }
 ```
 
@@ -804,6 +856,7 @@ EOF
 ## Task 3: Episode page as a static document
 
 **Files:**
+
 - Create: `src/components/EpisodeBody.tsx`, `src/routes/episodes.$slug.tsx`, `src/content/atproto.ts`
 - Modify: `scripts/check-build.mjs`
 - Test: `src/content/atproto.test.ts`
@@ -811,6 +864,7 @@ EOF
 **RENDER MECHANISM:** a server route handler at `/episodes/$slug` renders a complete HTML document with `renderToStaticMarkup` and returns it with `content-type: text/html`; the prerenderer writes it to `dist/client/episodes/<slug>/index.html`. The content appears once, there is no hydration, and the only script is an inline delegated click listener that seeks the native `<audio>` element.
 
 **Interfaces:**
+
 - Consumes: `loadEpisodes`, `getEpisode` (`src/content/load.ts`); `hms`, `toSeconds` (`src/content/time.ts`); `Episode`, `OutlineItem` (`src/content/types.ts`); `Document`, `SITE_NAME`, `SITE_URL` (`src/components/Document.tsx`).
 - Produces:
   - `<Outline items={OutlineItem[]} />` — exported from `EpisodeBody.tsx`, reused by Task 4.
@@ -821,11 +875,13 @@ EOF
   - Prerendered `dist/client/episodes/<slug>/index.html` for every episode.
 
 **AT Protocol additions on this page** (all optional — the page renders identically when the front matter and env vars are absent):
+
 - `<link rel="site.standard.document" href={episode.atUri}>` in the head when `atUri` is set.
 - After the transcript, when `bskyPostUrl` is set: a "Reply on Bluesky" link and `<section id="comments" data-thread="at://…">`. An inline script fetches the reply thread from the public Bluesky API and appends nested `<ol class="replies">`. **This is the only external request an episode page makes**, and any failure — no DID, a 404, an offline reader — leaves the section showing just the link.
 - The `at://` URI in `data-thread` is derived **at build time with no network call**: `at://<did>/app.bsky.feed.post/<rkey>`, where `<rkey>` is the last segment of `bskyPostUrl` and `<did>` is the profile segment when it already is a DID, otherwise `VITE_ATPROTO_DID`. Resolving a handle would mean a network call during prerender, which the build must not need.
 
 **Markup contract** (Tasks 4, 7, 9, 10 depend on it):
+
 - Outline is a nested `<ol class="outline">`; each timestamp is a `<button class="ts" data-seconds="30">`, each title an `<a href="#anchor">`, plus an outbound `<a class="outbound">` when `url` is set.
 - Sections are `<section>` with `<h2 id={anchor}>`.
 - Segments are `<p class="segment">` with `<strong class="speaker">` and `<time class="ts" data-seconds="30">`.
@@ -837,30 +893,30 @@ EOF
 In `scripts/check-build.mjs`, insert this block immediately before the `console.log('\nSizes')` line:
 
 ```js
-console.log('\nTask 3: episode documents')
-const episodeSlugs = slugs()
-check('content/episodes has at least one episode', episodeSlugs.length > 0)
-for (const slug of episodeSlugs) hasFile(`episodes/${slug}/index.html`)
-const newest = episodeSlugs.at(-1)
+console.log("\nTask 3: episode documents");
+const episodeSlugs = slugs();
+check("content/episodes has at least one episode", episodeSlugs.length > 0);
+for (const slug of episodeSlugs) hasFile(`episodes/${slug}/index.html`);
+const newest = episodeSlugs.at(-1);
 if (newest) {
-  const path = `episodes/${newest}/index.html`
-  contains(path, 'data-pagefind-body')
-  contains(path, 'class="outline"')
-  contains(path, 'data-seconds=')
-  contains(path, '<audio')
-  const html = file(path) || ''
+  const path = `episodes/${newest}/index.html`;
+  contains(path, "data-pagefind-body");
+  contains(path, 'class="outline"');
+  contains(path, "data-seconds=");
+  contains(path, "<audio");
+  const html = file(path) || "";
   check(
     `${path} references no bundled JS`,
     !/\/assets\/[^"']*\.js/.test(html),
-    'static documents must ship zero external scripts',
-  )
+    "static documents must ship zero external scripts",
+  );
   // AT Protocol bits are conditional on front matter and env, so only assert
   // them when the inputs are actually present.
-  const md = readFileSync(join('content/episodes', `${newest}.md`), 'utf8')
-  if (/^atUri:/m.test(md)) contains(path, 'rel="site.standard.document"')
+  const md = readFileSync(join("content/episodes", `${newest}.md`), "utf8");
+  if (/^atUri:/m.test(md)) contains(path, 'rel="site.standard.document"');
   if (/^bskyPostUrl:/m.test(md)) {
-    contains(path, 'Reply on Bluesky')
-    if (process.env.VITE_ATPROTO_DID) contains(path, 'data-thread="at://')
+    contains(path, "Reply on Bluesky");
+    if (process.env.VITE_ATPROTO_DID) contains(path, 'data-thread="at://');
   }
 }
 ```
@@ -875,34 +931,46 @@ Expected: FAIL. The build itself fails first at prerender, because `pages[]` lis
 Create `src/content/atproto.test.ts`:
 
 ```ts
-import test from 'node:test'
-import assert from 'node:assert/strict'
-import { bskyPostToAtUri } from './atproto.ts'
+import test from "node:test";
+import assert from "node:assert/strict";
+import { bskyPostToAtUri } from "./atproto.ts";
 
-const DID = 'did:plc:showaccount'
+const DID = "did:plc:showaccount";
 
-test('builds an at:// URI from a handle URL plus the configured DID', () => {
+test("builds an at:// URI from a handle URL plus the configured DID", () => {
   assert.equal(
-    bskyPostToAtUri('https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x', DID),
-    'at://did:plc:showaccount/app.bsky.feed.post/3lqz7abcd2k2x',
-  )
-})
-
-test('uses the DID already in the URL and ignores the configured one', () => {
+    bskyPostToAtUri(
+      "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
+      DID,
+    ),
+    "at://did:plc:showaccount/app.bsky.feed.post/3lqz7abcd2k2x",
+  );
+});
+
+test("uses the DID already in the URL and ignores the configured one", () => {
   assert.equal(
-    bskyPostToAtUri('https://bsky.app/profile/did:plc:other/post/3lqz7abcd2k2x', DID),
-    'at://did:plc:other/app.bsky.feed.post/3lqz7abcd2k2x',
-  )
-})
-
-test('returns undefined without a DID, or for a non-post URL', () => {
+    bskyPostToAtUri(
+      "https://bsky.app/profile/did:plc:other/post/3lqz7abcd2k2x",
+      DID,
+    ),
+    "at://did:plc:other/app.bsky.feed.post/3lqz7abcd2k2x",
+  );
+});
+
+test("returns undefined without a DID, or for a non-post URL", () => {
   assert.equal(
-    bskyPostToAtUri('https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x', undefined),
+    bskyPostToAtUri(
+      "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
+      undefined,
+    ),
     undefined,
-  )
-  assert.equal(bskyPostToAtUri('https://bsky.app/profile/thismonthinreact.com', DID), undefined)
-  assert.equal(bskyPostToAtUri('https://example.com/whatever', DID), undefined)
-})
+  );
+  assert.equal(
+    bskyPostToAtUri("https://bsky.app/profile/thismonthinreact.com", DID),
+    undefined,
+  );
+  assert.equal(bskyPostToAtUri("https://example.com/whatever", DID), undefined);
+});
 ```
 
 Run: `npm test`
@@ -919,12 +987,17 @@ Pure, no `node:` imports and no `import.meta.env`, so both a component and `node
  * caller passes the show account's DID (VITE_ATPROTO_DID). Returns undefined
  * when the URL is not a post URL, or when a handle URL has no DID to pair with.
  */
-export function bskyPostToAtUri(url: string, did: string | undefined): string | undefined {
-  const match = /^https:\/\/bsky\.app\/profile\/([^/]+)\/post\/([^/?#]+)/.exec(url.trim())
-  if (!match) return undefined
-  const repo = match[1].startsWith('did:') ? match[1] : did
-  if (!repo) return undefined
-  return `at://${repo}/app.bsky.feed.post/${match[2]}`
+export function bskyPostToAtUri(
+  url: string,
+  did: string | undefined,
+): string | undefined {
+  const match = /^https:\/\/bsky\.app\/profile\/([^/]+)\/post\/([^/?#]+)/.exec(
+    url.trim(),
+  );
+  if (!match) return undefined;
+  const repo = match[1].startsWith("did:") ? match[1] : did;
+  if (!repo) return undefined;
+  return `at://${repo}/app.bsky.feed.post/${match[2]}`;
 }
 ```
 
@@ -936,15 +1009,15 @@ Expected: PASS.
 A plain component — no `'use client'`, no hooks, no event handlers. `renderToStaticMarkup` will render it, so anything interactive here would be silently dead.
 
 ```tsx
-import type { Episode, OutlineItem } from '../content/types.ts'
-import { hms, toSeconds } from '../content/time.ts'
+import type { Episode, OutlineItem } from "../content/types.ts";
+import { hms, toSeconds } from "../content/time.ts";
 
 /** One delegated listener seeks the page's single <audio> from any [data-seconds]. */
 export const SEEK_SCRIPT = `document.addEventListener("click",function(e){
 var t=e.target.closest("[data-seconds]");if(!t)return;
 var a=document.querySelector("audio");if(!a)return;
 var s=Number(t.dataset.seconds);if(!isFinite(s))return;
-e.preventDefault();a.currentTime=s;a.play()});`
+e.preventDefault();a.currentTime=s;a.play()});`;
 
 /**
  * Fetches the announcement post's reply thread from the public Bluesky API and
@@ -984,26 +1057,26 @@ fetch("https://public.api.bsky.app/xrpc/app.bsky.feed.getPostThread?uri="+encode
   var list=render(t.replies);if(list)el.appendChild(list);
 })
 .catch(function(){});
-})();`
+})();`;
 
 export function Outline({ items }: { items: OutlineItem[] }) {
   return (
     <ol className="outline">
       {items.map((item, i) => {
-        const seconds = toSeconds(item.time)
+        const seconds = toSeconds(item.time);
         return (
           <li key={`${item.anchor}-${i}`}>
             {seconds !== undefined && (
               <>
                 <button type="button" className="ts" data-seconds={seconds}>
                   {hms(seconds)}
-                </button>{' '}
+                </button>{" "}
               </>
             )}
             <a href={`#${item.anchor}`}>{item.title}</a>
             {item.url && (
               <>
-                {' '}
+                {" "}
                 <a className="outbound" href={item.url} rel="noreferrer">
                   &#8599;
                 </a>
@@ -1011,10 +1084,10 @@ export function Outline({ items }: { items: OutlineItem[] }) {
             )}
             {item.children.length > 0 && <Outline items={item.children} />}
           </li>
-        )
+        );
       })}
     </ol>
-  )
+  );
 }
 
 export function EpisodeBody({ episode }: { episode: Episode }) {
@@ -1029,27 +1102,29 @@ export function EpisodeBody({ episode }: { episode: Episode }) {
           <section key={section.anchor}>
             <h2 id={section.anchor}>{section.title}</h2>
             {section.segments.map((segment, i) => {
-              const seconds = toSeconds(segment.time)
+              const seconds = toSeconds(segment.time);
               return (
                 <p className="segment" key={i}>
-                  {segment.speaker && <strong className="speaker">{segment.speaker}: </strong>}
+                  {segment.speaker && (
+                    <strong className="speaker">{segment.speaker}: </strong>
+                  )}
                   {segment.text}
                   {seconds !== undefined && (
                     <>
-                      {' '}
+                      {" "}
                       <time className="ts" data-seconds={seconds}>
                         {hms(seconds)}
                       </time>
                     </>
                   )}
                 </p>
-              )
+              );
             })}
           </section>
         ))}
       </div>
     </div>
-  )
+  );
 }
 ```
 
@@ -1058,26 +1133,30 @@ export function EpisodeBody({ episode }: { episode: Episode }) {
 This file has a `server.handlers.GET` and **no `component`** — the two cannot coexist on one path.
 
 ```tsx
-import { createFileRoute } from '@tanstack/react-router'
-import { renderToStaticMarkup } from 'react-dom/server'
-import { Document, SITE_NAME, SITE_URL } from '../components/Document'
-import { COMMENTS_SCRIPT, EpisodeBody, SEEK_SCRIPT } from '../components/EpisodeBody'
-import { bskyPostToAtUri } from '../content/atproto.ts'
-
-const ATPROTO_DID = import.meta.env.VITE_ATPROTO_DID
-
-export const Route = createFileRoute('/episodes/$slug')({
+import { createFileRoute } from "@tanstack/react-router";
+import { renderToStaticMarkup } from "react-dom/server";
+import { Document, SITE_NAME, SITE_URL } from "../components/Document";
+import {
+  COMMENTS_SCRIPT,
+  EpisodeBody,
+  SEEK_SCRIPT,
+} from "../components/EpisodeBody";
+import { bskyPostToAtUri } from "../content/atproto.ts";
+
+const ATPROTO_DID = import.meta.env.VITE_ATPROTO_DID;
+
+export const Route = createFileRoute("/episodes/$slug")({
   server: {
     handlers: {
       GET: async ({ params }) => {
-        const { getEpisode } = await import('../content/load.ts')
-        const episode = await getEpisode(params.slug)
-        if (!episode) return new Response('Not found', { status: 404 })
+        const { getEpisode } = await import("../content/load.ts");
+        const episode = await getEpisode(params.slug);
+        if (!episode) return new Response("Not found", { status: 404 });
 
         // Derived at build time; no network call during prerender.
         const threadUri = episode.bskyPostUrl
           ? bskyPostToAtUri(episode.bskyPostUrl, ATPROTO_DID)
-          : undefined
+          : undefined;
 
         const html = renderToStaticMarkup(
           <Document
@@ -1085,15 +1164,24 @@ export const Route = createFileRoute('/episodes/$slug')({
               <>
                 <title>{`${episode.title} — ${SITE_NAME}`}</title>
                 <meta name="description" content={episode.description} />
-                <link rel="canonical" href={`${SITE_URL}/episodes/${episode.slug}`} />
-                {episode.atUri && <link rel="site.standard.document" href={episode.atUri} />}
+                <link
+                  rel="canonical"
+                  href={`${SITE_URL}/episodes/${episode.slug}`}
+                />
+                {episode.atUri && (
+                  <link rel="site.standard.document" href={episode.atUri} />
+                )}
               </>
             }
             bodyClass="episode-page"
             scripts={
               <>
                 <script dangerouslySetInnerHTML={{ __html: SEEK_SCRIPT }} />
-                {threadUri && <script dangerouslySetInnerHTML={{ __html: COMMENTS_SCRIPT }} />}
+                {threadUri && (
+                  <script
+                    dangerouslySetInnerHTML={{ __html: COMMENTS_SCRIPT }}
+                  />
+                )}
               </>
             }
           >
@@ -1105,7 +1193,11 @@ export const Route = createFileRoute('/episodes/$slug')({
             )}
             <EpisodeBody episode={episode} />
             {episode.bskyPostUrl && (
-              <section id="comments" data-thread={threadUri} data-pagefind-ignore="">
+              <section
+                id="comments"
+                data-thread={threadUri}
+                data-pagefind-ignore=""
+              >
                 <h2>Comments</h2>
                 <p>
                   <a href={episode.bskyPostUrl} rel="noreferrer">
@@ -1115,15 +1207,15 @@ export const Route = createFileRoute('/episodes/$slug')({
               </section>
             )}
           </Document>,
-        )
+        );
 
         return new Response(`<!DOCTYPE html>${html}`, {
-          headers: { 'content-type': 'text/html; charset=utf-8' },
-        })
+          headers: { "content-type": "text/html; charset=utf-8" },
+        });
       },
     },
   },
-})
+});
 ```
 
 - [ ] **Step 5: Run build and check to verify they pass**
@@ -1134,9 +1226,11 @@ Expected: build prerenders one document per episode; all Task 3 lines `ok`, incl
 - [ ] **Step 6: Confirm the content is not duplicated**
 
 Run (substitute a real slug and a distinctive phrase from that episode's transcript):
+
 ```bash
 grep -o 'A_PHRASE_FROM_THE_TRANSCRIPT' dist/client/episodes/<slug>/index.html | wc -l
 ```
+
 Expected: `1`. If it is `2`, something has reintroduced a loader or RSC payload — stop and fix before continuing.
 
 - [ ] **Step 7: Commit**
@@ -1159,12 +1253,14 @@ EOF
 ## Task 4: Home route
 
 **Files:**
+
 - Create: `src/components/Player.tsx`
 - Modify: `src/routes/index.tsx`, `scripts/check-build.mjs`
 
 **RENDER MECHANISM:** `/` stays a router page. Its loader returns plain serializable data (the latest episode's outline plus a slim archive list) — no RSC, because a duplicated outline is a few kilobytes rather than a duplicated transcript. The player is a `'use client'` component.
 
 **Interfaces:**
+
 - Consumes: `loadEpisodes` (`src/content/load.ts`); `Outline` (`src/components/EpisodeBody.tsx`); `SITE_NAME`, `SITE_URL` (`src/components/Document.tsx`); `OutlineItem` (`src/content/types.ts`).
 - Produces:
   - `<Player audioUrl={string | undefined} title={string} />` — `'use client'`, native `<audio>` plus the same delegated `[data-seconds]` seek behaviour, as a React effect.
@@ -1175,9 +1271,9 @@ EOF
 In `scripts/check-build.mjs`, add immediately after the `contains('index.html', 'href="/styles.css"')` line:
 
 ```js
-contains('index.html', 'class="archive"')
-contains('index.html', 'class="outline"')
-for (const slug of slugs()) contains('index.html', `/episodes/${slug}`)
+contains("index.html", 'class="archive"');
+contains("index.html", 'class="outline"');
+for (const slug of slugs()) contains("index.html", `/episodes/${slug}`);
 ```
 
 - [ ] **Step 2: Run build and check to verify it fails**
@@ -1188,33 +1284,47 @@ Expected: FAIL — `index.html contains "class=\"archive\""`.
 - [ ] **Step 3: Write `src/components/Player.tsx`**
 
 ```tsx
-'use client'
-import { useEffect, useRef } from 'react'
+"use client";
+import { useEffect, useRef } from "react";
 
-export function Player({ audioUrl, title }: { audioUrl?: string; title: string }) {
-  const ref = useRef<HTMLAudioElement>(null)
+export function Player({
+  audioUrl,
+  title,
+}: {
+  audioUrl?: string;
+  title: string;
+}) {
+  const ref = useRef<HTMLAudioElement>(null);
 
   useEffect(() => {
     function onClick(event: MouseEvent) {
-      const target = (event.target as HTMLElement | null)?.closest<HTMLElement>('[data-seconds]')
-      const audio = ref.current
-      if (!target || !audio) return
-      const seconds = Number(target.dataset.seconds)
-      if (!Number.isFinite(seconds)) return
-      event.preventDefault()
-      audio.currentTime = seconds
-      void audio.play()
+      const target = (event.target as HTMLElement | null)?.closest<HTMLElement>(
+        "[data-seconds]",
+      );
+      const audio = ref.current;
+      if (!target || !audio) return;
+      const seconds = Number(target.dataset.seconds);
+      if (!Number.isFinite(seconds)) return;
+      event.preventDefault();
+      audio.currentTime = seconds;
+      void audio.play();
     }
-    document.addEventListener('click', onClick)
-    return () => document.removeEventListener('click', onClick)
-  }, [])
+    document.addEventListener("click", onClick);
+    return () => document.removeEventListener("click", onClick);
+  }, []);
 
-  if (!audioUrl) return null
+  if (!audioUrl) return null;
   return (
     <div className="player" data-pagefind-ignore="">
-      <audio ref={ref} controls preload="none" src={audioUrl} aria-label={`Play ${title}`} />
+      <audio
+        ref={ref}
+        controls
+        preload="none"
+        src={audioUrl}
+        aria-label={`Play ${title}`}
+      />
     </div>
-  )
+  );
 }
 ```
 
@@ -1223,17 +1333,17 @@ export function Player({ audioUrl, title }: { audioUrl?: string; title: string }
 Episode links are plain `<a href>`, not `<Link>`, because episode pages are outside the router.
 
 ```tsx
-import { createFileRoute } from '@tanstack/react-router'
-import { createServerFn } from '@tanstack/react-start'
-import { Outline } from '../components/EpisodeBody'
-import { Player } from '../components/Player'
-import { SITE_NAME, SITE_URL } from '../components/Document'
-import type { OutlineItem } from '../content/types.ts'
+import { createFileRoute } from "@tanstack/react-router";
+import { createServerFn } from "@tanstack/react-start";
+import { Outline } from "../components/EpisodeBody";
+import { Player } from "../components/Player";
+import { SITE_NAME, SITE_URL } from "../components/Document";
+import type { OutlineItem } from "../content/types.ts";
 
 const getHome = createServerFn().handler(async () => {
-  const { loadEpisodes } = await import('../content/load.ts')
-  const episodes = await loadEpisodes()
-  const latest = episodes[0]
+  const { loadEpisodes } = await import("../content/load.ts");
+  const episodes = await loadEpisodes();
+  const latest = episodes[0];
   return {
     latest: latest
       ? {
@@ -1249,26 +1359,26 @@ const getHome = createServerFn().handler(async () => {
       date: e.date,
       description: e.description,
     })),
-  }
-})
+  };
+});
 
-export const Route = createFileRoute('/')({
+export const Route = createFileRoute("/")({
   loader: () => getHome(),
   head: () => ({
     meta: [
       { title: SITE_NAME },
       {
-        name: 'description',
+        name: "description",
         content: `Every episode of ${SITE_NAME}, with transcripts and links.`,
       },
     ],
-    links: [{ rel: 'canonical', href: `${SITE_URL}/` }],
+    links: [{ rel: "canonical", href: `${SITE_URL}/` }],
   }),
   component: Home,
-})
+});
 
 function Home() {
-  const { latest, archive } = Route.useLoaderData()
+  const { latest, archive } = Route.useLoaderData();
   return (
     <>
       {latest && (
@@ -1293,7 +1403,7 @@ function Home() {
         </ul>
       </section>
     </>
-  )
+  );
 }
 ```
 
@@ -1321,12 +1431,14 @@ EOF
 ## Task 5: Link index as a static document
 
 **Files:**
+
 - Create: `src/routes/links.tsx`
 - Modify: `src/components/Document.tsx`, `scripts/check-build.mjs`
 
 **RENDER MECHANISM:** identical to Task 3 — a server route handler renders a full document with `renderToStaticMarkup`, prerendered to `dist/client/links/index.html`. The filter is an inline vanilla script that toggles `hidden` on already-rendered nodes; the list is never re-rendered and no framework code loads.
 
 **Interfaces:**
+
 - Consumes: `loadEpisodes` (`src/content/load.ts`); `buildLinkIndex`, `groupByHost`, `LinkEntry` (`src/content/links.ts`); `hms` (`src/content/time.ts`); `Document`, `SITE_NAME`, `SITE_URL`.
 - Produces: `dist/client/links/index.html`. Every row is `<li class="link-row" data-text="…">` where `data-text` is the lowercased link text plus hostname; groups are `<section class="link-group">`; the input is `<input id="link-filter" type="search">`.
 
@@ -1335,15 +1447,15 @@ EOF
 Append to `scripts/check-build.mjs`, immediately before the `console.log('\nSizes')` line:
 
 ```js
-console.log('\nTask 5: links document')
-hasFile('links/index.html')
-contains('links/index.html', 'class="link-row"')
-contains('links/index.html', 'data-text=')
-contains('links/index.html', 'id="link-filter"')
+console.log("\nTask 5: links document");
+hasFile("links/index.html");
+contains("links/index.html", 'class="link-row"');
+contains("links/index.html", "data-text=");
+contains("links/index.html", 'id="link-filter"');
 check(
-  'links/index.html references no bundled JS',
-  !/\/assets\/[^"']*\.js/.test(file('links/index.html') || ''),
-)
+  "links/index.html references no bundled JS",
+  !/\/assets\/[^"']*\.js/.test(file("links/index.html") || ""),
+);
 ```
 
 - [ ] **Step 2: Run build and check to verify it fails**
@@ -1354,9 +1466,9 @@ Expected: FAIL — `links/index.html exists`.
 - [ ] **Step 3: Write `src/routes/links.tsx`**
 
 ```tsx
-import { createFileRoute } from '@tanstack/react-router'
-import { renderToStaticMarkup } from 'react-dom/server'
-import { Document, SITE_NAME, SITE_URL } from '../components/Document'
+import { createFileRoute } from "@tanstack/react-router";
+import { renderToStaticMarkup } from "react-dom/server";
+import { Document, SITE_NAME, SITE_URL } from "../components/Document";
 
 const FILTER_SCRIPT = `(function(){
 var input=document.getElementById("link-filter");if(!input)return;
@@ -1366,16 +1478,17 @@ input.addEventListener("input",function(){
 var q=input.value.trim().toLowerCase();
 for(var i=0;i<rows.length;i++){rows[i].hidden=q!==""&&rows[i].dataset.text.indexOf(q)===-1;}
 for(var j=0;j<groups.length;j++){groups[j].hidden=!groups[j].querySelector(".link-row:not([hidden])");}
-});})();`
+});})();`;
 
-export const Route = createFileRoute('/links')({
+export const Route = createFileRoute("/links")({
   server: {
     handlers: {
       GET: async () => {
-        const { loadEpisodes } = await import('../content/load.ts')
-        const { buildLinkIndex, groupByHost } = await import('../content/links.ts')
-        const { hms } = await import('../content/time.ts')
-        const groups = groupByHost(buildLinkIndex(await loadEpisodes()))
+        const { loadEpisodes } = await import("../content/load.ts");
+        const { buildLinkIndex, groupByHost } =
+          await import("../content/links.ts");
+        const { hms } = await import("../content/time.ts");
+        const groups = groupByHost(buildLinkIndex(await loadEpisodes()));
 
         const html = renderToStaticMarkup(
           <Document
@@ -1390,7 +1503,9 @@ export const Route = createFileRoute('/links')({
               </>
             }
             bodyClass="links-page"
-            scripts={<script dangerouslySetInnerHTML={{ __html: FILTER_SCRIPT }} />}
+            scripts={
+              <script dangerouslySetInnerHTML={{ __html: FILTER_SCRIPT }} />
+            }
           >
             <h1>Links</h1>
             <input
@@ -1402,7 +1517,8 @@ export const Route = createFileRoute('/links')({
             {groups.map((group) => (
               <section className="link-group" key={group.host}>
                 <h2>
-                  {group.host} <span className="count">{group.entries.length}</span>
+                  {group.host}{" "}
+                  <span className="count">{group.entries.length}</span>
                 </h2>
                 <ul>
                   {group.entries.map((entry, i) => (
@@ -1414,7 +1530,10 @@ export const Route = createFileRoute('/links')({
                       <a href={entry.url} rel="noreferrer">
                         {entry.text}
                       </a>
-                      <a className="from" href={`/episodes/${entry.episodeSlug}`}>
+                      <a
+                        className="from"
+                        href={`/episodes/${entry.episodeSlug}`}
+                      >
                         {entry.episodeTitle}
                         {entry.time !== undefined && (
                           <span className="ts">{hms(entry.time)}</span>
@@ -1426,15 +1545,15 @@ export const Route = createFileRoute('/links')({
               </section>
             ))}
           </Document>,
-        )
+        );
 
         return new Response(`<!DOCTYPE html>${html}`, {
-          headers: { 'content-type': 'text/html; charset=utf-8' },
-        })
+          headers: { "content-type": "text/html; charset=utf-8" },
+        });
       },
     },
   },
-})
+});
 ```
 
 > The `.from` link goes to the top of the episode page rather than to a fragment: a link entry's anchor is not guaranteed to be a section heading id.
@@ -1470,11 +1589,13 @@ EOF
 ## Task 6: RSS feed and chapters JSON
 
 **Files:**
+
 - Create: `src/content/feed.ts`, `src/routes/feed[.]xml.ts`, `src/routes/episodes.$slug.chapters[.]json.ts`
 - Test: `src/content/feed.test.ts`
 - Modify: `scripts/check-build.mjs`
 
 **Interfaces:**
+
 - Consumes: `loadEpisodes`, `getEpisode` (`src/content/load.ts`); `toSeconds` (`src/content/time.ts`); `Episode`, `OutlineItem` (types); `SITE_NAME`, `SITE_URL`.
 - Produces:
   - `outlineToHtml(items: OutlineItem[]): string`
@@ -1489,50 +1610,55 @@ EOF
 `src/content/feed.test.ts`:
 
 ```ts
-import { test } from 'node:test'
-import assert from 'node:assert/strict'
-
-process.env.TMIR_CONTENT_DIR = 'test/fixtures/episodes'
-const { outlineToHtml, outlineToChapters, renderFeed } = await import('./feed.ts')
-const { loadEpisodes } = await import('./load.ts')
-
-test('outlineToHtml nests lists and links out', async () => {
-  const [episode] = await loadEpisodes()
-  const html = outlineToHtml(episode.outline)
-  assert.match(html, /^<ul>/)
-  assert.ok(html.includes('<a href="https://react.dev/compiler">React Compiler</a>'))
-  assert.ok(html.includes('<ul><li>'), 'nested child list present')
-  assert.ok(html.includes('No link here'))
-})
-
-test('outlineToHtml escapes markup in titles', () => {
-  const html = outlineToHtml([{ title: 'a & b <script>', anchor: 'a', children: [] }])
-  assert.ok(html.includes('a &amp; b &lt;script&gt;'))
-  assert.ok(!html.includes('<script>'))
-})
-
-test('outlineToChapters flattens timed items in order', async () => {
-  const [episode] = await loadEpisodes()
-  const chapters = outlineToChapters(episode.outline)
+import { test } from "node:test";
+import assert from "node:assert/strict";
+
+process.env.TMIR_CONTENT_DIR = "test/fixtures/episodes";
+const { outlineToHtml, outlineToChapters, renderFeed } =
+  await import("./feed.ts");
+const { loadEpisodes } = await import("./load.ts");
+
+test("outlineToHtml nests lists and links out", async () => {
+  const [episode] = await loadEpisodes();
+  const html = outlineToHtml(episode.outline);
+  assert.match(html, /^<ul>/);
+  assert.ok(
+    html.includes('<a href="https://react.dev/compiler">React Compiler</a>'),
+  );
+  assert.ok(html.includes("<ul><li>"), "nested child list present");
+  assert.ok(html.includes("No link here"));
+});
+
+test("outlineToHtml escapes markup in titles", () => {
+  const html = outlineToHtml([
+    { title: "a & b <script>", anchor: "a", children: [] },
+  ]);
+  assert.ok(html.includes("a &amp; b &lt;script&gt;"));
+  assert.ok(!html.includes("<script>"));
+});
+
+test("outlineToChapters flattens timed items in order", async () => {
+  const [episode] = await loadEpisodes();
+  const chapters = outlineToChapters(episode.outline);
   assert.deepEqual(
     chapters.map((c) => [c.startTime, c.title]),
     [
-      [30, 'React Compiler'],
-      [130, 'Caveats'],
-      [600, 'TanStack Start'],
+      [30, "React Compiler"],
+      [130, "Caveats"],
+      [600, "TanStack Start"],
     ],
-  )
-  assert.equal(chapters[0].url, 'https://react.dev/compiler')
-})
-
-test('renderFeed produces one item per episode with the outline as body', async () => {
-  const xml = renderFeed(await loadEpisodes(), 'TMiR', 'https://example.com')
-  assert.match(xml, /^<\?xml version="1\.0" encoding="UTF-8"\?>/)
-  assert.equal(xml.match(/<item>/g)?.length, 1)
-  assert.ok(xml.includes('<link>https://example.com/episodes/2026-05</link>'))
-  assert.ok(xml.includes('React Compiler'))
-  assert.ok(xml.includes('<![CDATA['))
-})
+  );
+  assert.equal(chapters[0].url, "https://react.dev/compiler");
+});
+
+test("renderFeed produces one item per episode with the outline as body", async () => {
+  const xml = renderFeed(await loadEpisodes(), "TMiR", "https://example.com");
+  assert.match(xml, /^<\?xml version="1\.0" encoding="UTF-8"\?>/);
+  assert.equal(xml.match(/<item>/g)?.length, 1);
+  assert.ok(xml.includes("<link>https://example.com/episodes/2026-05</link>"));
+  assert.ok(xml.includes("React Compiler"));
+  assert.ok(xml.includes("<![CDATA["));
+});
 ```
 
 - [ ] **Step 2: Run tests to verify they fail**
@@ -1545,64 +1671,70 @@ Expected: FAIL — `Cannot find module './feed.ts'`.
 Hand-rolled XML: a feed library would be a new dependency for forty lines of string building.
 
 ```ts
-import type { Episode, OutlineItem } from './types.ts'
-import { toSeconds } from './time.ts'
+import type { Episode, OutlineItem } from "./types.ts";
+import { toSeconds } from "./time.ts";
 
 function escapeHtml(text: string): string {
   return text
-    .replace(/&/g, '&amp;')
-    .replace(/</g, '&lt;')
-    .replace(/>/g, '&gt;')
-    .replace(/"/g, '&quot;')
+    .replace(/&/g, "&amp;")
+    .replace(/</g, "&lt;")
+    .replace(/>/g, "&gt;")
+    .replace(/"/g, "&quot;");
 }
 
 export function outlineToHtml(items: OutlineItem[]): string {
-  if (items.length === 0) return ''
+  if (items.length === 0) return "";
   const lis = items.map((item) => {
-    const title = escapeHtml(item.title)
-    const label = item.url ? `<a href="${escapeHtml(item.url)}">${title}</a>` : title
-    return `<li>${label}${outlineToHtml(item.children)}</li>`
-  })
-  return `<ul>${lis.join('')}</ul>`
+    const title = escapeHtml(item.title);
+    const label = item.url
+      ? `<a href="${escapeHtml(item.url)}">${title}</a>`
+      : title;
+    return `<li>${label}${outlineToHtml(item.children)}</li>`;
+  });
+  return `<ul>${lis.join("")}</ul>`;
 }
 
 export function outlineToChapters(
   items: OutlineItem[],
 ): { startTime: number; title: string; url?: string }[] {
-  const out: { startTime: number; title: string; url?: string }[] = []
+  const out: { startTime: number; title: string; url?: string }[] = [];
   const walk = (list: OutlineItem[]) => {
     for (const item of list) {
-      const startTime = toSeconds(item.time)
+      const startTime = toSeconds(item.time);
       if (startTime !== undefined) {
         out.push(
           item.url
             ? { startTime, title: item.title, url: item.url }
             : { startTime, title: item.title },
-        )
+        );
       }
-      walk(item.children)
+      walk(item.children);
     }
-  }
-  walk(items)
-  return out.sort((a, b) => a.startTime - b.startTime)
+  };
+  walk(items);
+  return out.sort((a, b) => a.startTime - b.startTime);
 }
 
-export function renderFeed(episodes: Episode[], siteName: string, siteUrl: string): string {
+export function renderFeed(
+  episodes: Episode[],
+  siteName: string,
+  siteUrl: string,
+): string {
   const items = episodes
     .map((episode) => {
-      const url = `${siteUrl}/episodes/${episode.slug}`
-      const body = `<p>${escapeHtml(episode.description)}</p>${outlineToHtml(episode.outline)}`
+      const url = `${siteUrl}/episodes/${episode.slug}`;
+      const body = `<p>${escapeHtml(episode.description)}</p>${outlineToHtml(episode.outline)}`;
       return [
-        '<item>',
+        "<item>",
         `<title>${escapeHtml(episode.title)}</title>`,
         `<link>${escapeHtml(url)}</link>`,
         `<guid isPermaLink="true">${escapeHtml(url)}</guid>`,
         `<pubDate>${new Date(episode.date).toUTCString()}</pubDate>`,
         `<description><![CDATA[${body}]]></description>`,
-        '</item>',
-      ].join('')
+        "</item>",
+      ].join("");
     })
-    .join('')
+    .join("");
 
   return [
     '<?xml version="1.0" encoding="UTF-8"?>',
@@ -1612,8 +1744,8 @@ export function renderFeed(episodes: Episode[], siteName: string, siteUrl: strin
     `<description>${escapeHtml(siteName)} episodes, outlines and links.</description>`,
     `<atom:link href="${escapeHtml(siteUrl)}/feed.xml" rel="self" type="application/rss+xml"/>`,
     items,
-    '</channel></rss>',
-  ].join('')
+    "</channel></rss>",
+  ].join("");
 }
 ```
 
@@ -1627,24 +1759,24 @@ Expected: PASS, 4 tests.
 Append to `scripts/check-build.mjs`, immediately before the `console.log('\nSizes')` line:
 
 ```js
-console.log('\nTask 6: feed and chapters')
-hasFile('feed.xml')
-contains('feed.xml', '<rss version="2.0"')
-contains('feed.xml', '<item>')
+console.log("\nTask 6: feed and chapters");
+hasFile("feed.xml");
+contains("feed.xml", '<rss version="2.0"');
+contains("feed.xml", "<item>");
 for (const slug of slugs()) {
-  const body = file(`episodes/${slug}/chapters.json`)
+  const body = file(`episodes/${slug}/chapters.json`);
   check(
     `episodes/${slug}/chapters.json is valid Podcasting 2.0`,
     (() => {
-      if (body === null) return false
+      if (body === null) return false;
       try {
-        const json = JSON.parse(body)
-        return json.version === '1.2.0' && Array.isArray(json.chapters)
+        const json = JSON.parse(body);
+        return json.version === "1.2.0" && Array.isArray(json.chapters);
       } catch {
-        return false
+        return false;
       }
     })(),
-  )
+  );
 }
 ```
 
@@ -1656,23 +1788,23 @@ Expected: FAIL — the build errors at prerender on the `/feed.xml` entry in `pa
 - [ ] **Step 7: Write `src/routes/feed[.]xml.ts`**
 
 ```ts
-import { createFileRoute } from '@tanstack/react-router'
-import { SITE_NAME, SITE_URL } from '../components/Document'
+import { createFileRoute } from "@tanstack/react-router";
+import { SITE_NAME, SITE_URL } from "../components/Document";
 
-export const Route = createFileRoute('/feed[.]xml')({
+export const Route = createFileRoute("/feed[.]xml")({
   server: {
     handlers: {
       GET: async () => {
-        const { loadEpisodes } = await import('../content/load.ts')
-        const { renderFeed } = await import('../content/feed.ts')
-        const xml = renderFeed(await loadEpisodes(), SITE_NAME, SITE_URL)
+        const { loadEpisodes } = await import("../content/load.ts");
+        const { renderFeed } = await import("../content/feed.ts");
+        const xml = renderFeed(await loadEpisodes(), SITE_NAME, SITE_URL);
         return new Response(xml, {
-          headers: { 'content-type': 'application/rss+xml; charset=utf-8' },
-        })
+          headers: { "content-type": "application/rss+xml; charset=utf-8" },
+        });
       },
     },
   },
-})
+});
 ```
 
 > Importing `SITE_NAME` / `SITE_URL` from `Document.tsx` pulls a `.tsx` module into a `.ts` file; that is fine, but if the bundler objects, inline the two `import.meta.env` reads here instead.
@@ -1680,24 +1812,24 @@ export const Route = createFileRoute('/feed[.]xml')({
 - [ ] **Step 8: Write `src/routes/episodes.$slug.chapters[.]json.ts`**
 
 ```ts
-import { createFileRoute } from '@tanstack/react-router'
+import { createFileRoute } from "@tanstack/react-router";
 
-export const Route = createFileRoute('/episodes/$slug/chapters[.]json')({
+export const Route = createFileRoute("/episodes/$slug/chapters[.]json")({
   server: {
     handlers: {
       GET: async ({ params }) => {
-        const { getEpisode } = await import('../content/load.ts')
-        const { outlineToChapters } = await import('../content/feed.ts')
-        const episode = await getEpisode(params.slug)
-        if (!episode) return new Response('Not found', { status: 404 })
+        const { getEpisode } = await import("../content/load.ts");
+        const { outlineToChapters } = await import("../content/feed.ts");
+        const episode = await getEpisode(params.slug);
+        if (!episode) return new Response("Not found", { status: 404 });
         return Response.json({
-          version: '1.2.0',
+          version: "1.2.0",
           chapters: outlineToChapters(episode.outline),
-        })
+        });
       },
     },
   },
-})
+});
 ```
 
 - [ ] **Step 9: Run build and check to verify they pass**
@@ -1722,10 +1854,12 @@ EOF
 ## Task 7: Search route
 
 **Files:**
+
 - Create: `src/routes/search.tsx`, `src/components/PagefindUI.tsx`
 - Modify: `src/components/Document.tsx`, `scripts/check-build.mjs`
 
 **Interfaces:**
+
 - Consumes: `SITE_NAME`, `SITE_URL`; the `data-pagefind-body` / `data-pagefind-ignore` attributes set in Tasks 1, 3 and 4.
 - Produces: `dist/client/search/index.html` and, after `npx pagefind --site dist/client`, `dist/client/pagefind/`.
 
@@ -1736,14 +1870,14 @@ Pagefind's assets do not exist when Vite builds, so they cannot be imported. `<P
 Append to `scripts/check-build.mjs`, immediately before the `console.log('\nSizes')` line:
 
 ```js
-console.log('\nTask 7: search')
-hasFile('search/index.html')
-contains('search/index.html', 'id="pagefind-ui"')
+console.log("\nTask 7: search");
+hasFile("search/index.html");
+contains("search/index.html", 'id="pagefind-ui"');
 check(
-  'pagefind index built',
-  existsSync(join(dist, 'pagefind', 'pagefind-ui.js')),
-  'run: npx pagefind --site dist/client',
-)
+  "pagefind index built",
+  existsSync(join(dist, "pagefind", "pagefind-ui.js")),
+  "run: npx pagefind --site dist/client",
+);
 ```
 
 - [ ] **Step 2: Run build and check to verify it fails**
@@ -1754,55 +1888,64 @@ Expected: FAIL — `search/index.html exists` and `pagefind index built`.
 - [ ] **Step 3: Write `src/components/PagefindUI.tsx`**
 
 ```tsx
-'use client'
-import { useEffect } from 'react'
+"use client";
+import { useEffect } from "react";
 
 declare global {
   interface Window {
-    PagefindUI?: new (options: { element: string; showSubResults?: boolean }) => unknown
+    PagefindUI?: new (options: {
+      element: string;
+      showSubResults?: boolean;
+    }) => unknown;
   }
 }
 
 export function PagefindUI() {
   useEffect(() => {
-    const css = document.createElement('link')
-    css.rel = 'stylesheet'
-    css.href = '/pagefind/pagefind-ui.css'
-    document.head.appendChild(css)
+    const css = document.createElement("link");
+    css.rel = "stylesheet";
+    css.href = "/pagefind/pagefind-ui.css";
+    document.head.appendChild(css);
 
-    const script = document.createElement('script')
-    script.src = '/pagefind/pagefind-ui.js'
+    const script = document.createElement("script");
+    script.src = "/pagefind/pagefind-ui.js";
     script.onload = () => {
       if (window.PagefindUI) {
-        new window.PagefindUI({ element: '#pagefind-ui', showSubResults: true })
+        new window.PagefindUI({
+          element: "#pagefind-ui",
+          showSubResults: true,
+        });
       }
-    }
-    document.head.appendChild(script)
+    };
+    document.head.appendChild(script);
 
     return () => {
-      css.remove()
-      script.remove()
-    }
-  }, [])
+      css.remove();
+      script.remove();
+    };
+  }, []);
 
-  return <div id="pagefind-ui" />
+  return <div id="pagefind-ui" />;
 }
 ```
 
 - [ ] **Step 4: Write `src/routes/search.tsx`**
 
 ```tsx
-import { createFileRoute } from '@tanstack/react-router'
-import { PagefindUI } from '../components/PagefindUI'
-import { SITE_NAME, SITE_URL } from '../components/Document'
+import { createFileRoute } from "@tanstack/react-router";
+import { PagefindUI } from "../components/PagefindUI";
+import { SITE_NAME, SITE_URL } from "../components/Document";
 
-export const Route = createFileRoute('/search')({
+export const Route = createFileRoute("/search")({
   head: () => ({
-    meta: [{ title: `Search — ${SITE_NAME}` }, { name: 'robots', content: 'noindex' }],
-    links: [{ rel: 'canonical', href: `${SITE_URL}/search` }],
+    meta: [
+      { title: `Search — ${SITE_NAME}` },
+      { name: "robots", content: "noindex" },
+    ],
+    links: [{ rel: "canonical", href: `${SITE_URL}/search` }],
   }),
   component: Search,
-})
+});
 
 function Search() {
   return (
@@ -1810,13 +1953,13 @@ function Search() {
       <h1>Search transcripts</h1>
       <noscript>
         <p>
-          Search needs JavaScript. Every transcript is on its episode page, and your
-          browser&rsquo;s find-in-page works there.
+          Search needs JavaScript. Every transcript is on its episode page, and
+          your browser&rsquo;s find-in-page works there.
         </p>
       </noscript>
       <PagefindUI />
     </section>
-  )
+  );
 }
 ```
 
@@ -1851,10 +1994,12 @@ EOF
 ## Task 8: About route
 
 **Files:**
+
 - Create: `src/routes/about.tsx`
 - Modify: `src/components/Document.tsx`, `scripts/check-build.mjs`
 
 **Interfaces:**
+
 - Consumes: `loadEpisodes` (`src/content/load.ts`); `Person` (types); `SITE_NAME`, `SITE_URL`.
 - Produces: `dist/client/about/index.html` with a people list, subscribe links, the live schedule from the latest episode, one sentence saying the site is an AT Protocol publication, a "Follow on Bluesky" link when `VITE_BLUESKY_PROFILE_URL` is set, and — only when `VITE_BUTTONDOWN_USER` is set — a Buttondown subscribe form.
 
@@ -1863,22 +2008,25 @@ EOF
 Append to `scripts/check-build.mjs`, immediately before the `console.log('\nSizes')` line:
 
 ```js
-console.log('\nTask 8: about')
-hasFile('about/index.html')
-contains('about/index.html', 'class="people"')
-contains('about/index.html', 'https://feeds.transistor.fm/this-month-in-react')
-contains('about/index.html', 'AT Protocol publication')
-if (process.env.VITE_BLUESKY_PROFILE_URL) contains('about/index.html', 'Follow on Bluesky')
-const about = file('about/index.html') || ''
-const bdUser = process.env.VITE_BUTTONDOWN_USER
+console.log("\nTask 8: about");
+hasFile("about/index.html");
+contains("about/index.html", 'class="people"');
+contains("about/index.html", "https://feeds.transistor.fm/this-month-in-react");
+contains("about/index.html", "AT Protocol publication");
+if (process.env.VITE_BLUESKY_PROFILE_URL)
+  contains("about/index.html", "Follow on Bluesky");
+const about = file("about/index.html") || "";
+const bdUser = process.env.VITE_BUTTONDOWN_USER;
 check(
   bdUser
-    ? 'newsletter form present when VITE_BUTTONDOWN_USER is set'
-    : 'newsletter form absent when VITE_BUTTONDOWN_USER is unset',
+    ? "newsletter form present when VITE_BUTTONDOWN_USER is set"
+    : "newsletter form absent when VITE_BUTTONDOWN_USER is unset",
   bdUser
-    ? about.includes(`https://buttondown.com/api/emails/embed-subscribe/${bdUser}`)
-    : !about.includes('buttondown.com'),
-)
+    ? about.includes(
+        `https://buttondown.com/api/emails/embed-subscribe/${bdUser}`,
+      )
+    : !about.includes("buttondown.com"),
+);
 ```
 
 - [ ] **Step 2: Run build and check to verify it fails**
@@ -1891,57 +2039,63 @@ Expected: FAIL — `about/index.html exists`.
 People are aggregated across every episode and deduplicated by name, keeping the first non-empty `role`, `href` and `img` seen.
 
 ```tsx
-import { createFileRoute } from '@tanstack/react-router'
-import { createServerFn } from '@tanstack/react-start'
-import { SITE_NAME, SITE_URL } from '../components/Document'
+import { createFileRoute } from "@tanstack/react-router";
+import { createServerFn } from "@tanstack/react-start";
+import { SITE_NAME, SITE_URL } from "../components/Document";
 
-const BUTTONDOWN_USER = import.meta.env.VITE_BUTTONDOWN_USER
-const BLUESKY_PROFILE_URL = import.meta.env.VITE_BLUESKY_PROFILE_URL
-const TRANSISTOR_FEED = 'https://feeds.transistor.fm/this-month-in-react'
+const BUTTONDOWN_USER = import.meta.env.VITE_BUTTONDOWN_USER;
+const BLUESKY_PROFILE_URL = import.meta.env.VITE_BLUESKY_PROFILE_URL;
+const TRANSISTOR_FEED = "https://feeds.transistor.fm/this-month-in-react";
 
 const getAbout = createServerFn().handler(async () => {
-  const { loadEpisodes } = await import('../content/load.ts')
-  const episodes = await loadEpisodes()
-  const byName = new Map<string, { name: string; role?: string; href?: string; img?: string }>()
+  const { loadEpisodes } = await import("../content/load.ts");
+  const episodes = await loadEpisodes();
+  const byName = new Map<
+    string,
+    { name: string; role?: string; href?: string; img?: string }
+  >();
   for (const episode of episodes) {
     for (const person of episode.people) {
-      const existing = byName.get(person.name)
+      const existing = byName.get(person.name);
       if (existing) {
-        existing.role ||= person.role
-        existing.href ||= person.href
-        existing.img ||= person.img
+        existing.role ||= person.role;
+        existing.href ||= person.href;
+        existing.img ||= person.img;
       } else {
-        byName.set(person.name, { ...person })
+        byName.set(person.name, { ...person });
       }
     }
   }
-  const latest = episodes[0]
+  const latest = episodes[0];
   return {
     people: [...byName.values()].sort((a, b) => a.name.localeCompare(b.name)),
     schedule: latest ? { time: latest.time, location: latest.location } : null,
-  }
-})
+  };
+});
 
-export const Route = createFileRoute('/about')({
+export const Route = createFileRoute("/about")({
   loader: () => getAbout(),
   head: () => ({
     meta: [
       { title: `About — ${SITE_NAME}` },
-      { name: 'description', content: `Who makes ${SITE_NAME}, and how to subscribe.` },
+      {
+        name: "description",
+        content: `Who makes ${SITE_NAME}, and how to subscribe.`,
+      },
     ],
-    links: [{ rel: 'canonical', href: `${SITE_URL}/about` }],
+    links: [{ rel: "canonical", href: `${SITE_URL}/about` }],
   }),
   component: About,
-})
+});
 
 function About() {
-  const { people, schedule } = Route.useLoaderData()
+  const { people, schedule } = Route.useLoaderData();
   return (
     <div className="about-page">
       <h1>About {SITE_NAME}</h1>
       <p>
-        A monthly news podcast about React and its ecosystem, hosted by Carl Vitullo and Mark
-        Erikson.
+        A monthly news podcast about React and its ecosystem, hosted by Carl
+        Vitullo and Mark Erikson.
       </p>
 
       {schedule?.time && (
@@ -1949,7 +2103,7 @@ function About() {
           <h2>Live</h2>
           <p>
             Recorded live at {schedule.time}
-            {schedule.location ? ` in ${schedule.location}` : ''}.
+            {schedule.location ? ` in ${schedule.location}` : ""}.
           </p>
         </section>
       )}
@@ -1959,7 +2113,15 @@ function About() {
         <ul>
           {people.map((person) => (
             <li key={person.name}>
-              {person.img && <img src={person.img} alt="" width="48" height="48" loading="lazy" />}
+              {person.img && (
+                <img
+                  src={person.img}
+                  alt=""
+                  width="48"
+                  height="48"
+                  loading="lazy"
+                />
+              )}
               {person.href ? (
                 <a href={person.href} rel="noreferrer">
                   {person.name}
@@ -1977,7 +2139,10 @@ function About() {
         <h2>Subscribe</h2>
         <ul>
           <li>
-            <a href="https://open.spotify.com/show/4g3Le83YfsMeI8Fq3cpPeH" rel="noreferrer">
+            <a
+              href="https://open.spotify.com/show/4g3Le83YfsMeI8Fq3cpPeH"
+              rel="noreferrer"
+            >
               Spotify
             </a>
           </li>
@@ -2003,8 +2168,9 @@ function About() {
           )}
         </ul>
         <p>
-          This site is an AT Protocol publication, so every episode is also a record on the
-          open network that Bluesky and other standard.site readers can follow.
+          This site is an AT Protocol publication, so every episode is also a
+          record on the open network that Bluesky and other standard.site
+          readers can follow.
         </p>
       </section>
 
@@ -2018,13 +2184,19 @@ function About() {
             target="_blank"
           >
             <label htmlFor="bd-email">Email</label>
-            <input id="bd-email" type="email" name="email" required autoComplete="email" />
+            <input
+              id="bd-email"
+              type="email"
+              name="email"
+              required
+              autoComplete="email"
+            />
             <button type="submit">Subscribe</button>
           </form>
         </section>
       )}
     </div>
-  )
+  );
 }
 ```
 
@@ -2062,9 +2234,11 @@ EOF
 ## Task 9: Stylesheet
 
 **Files:**
+
 - Modify: `public/styles.css`, `scripts/check-build.mjs`
 
 **Interfaces:**
+
 - Consumes: the class names emitted in Tasks 1 and 3–8 — `.site-header`, `.site-name`, `.site-nav`, `.site-footer`, `.episode-page-main`, `.episode`, `.toc`, `.outline`, `.transcript`, `.segment`, `.speaker`, `.ts`, `.outbound`, `.player`, `.latest`, `.archive`, `.links-page-main`, `.link-group`, `.link-row`, `.count`, `.from`, `#link-filter`, `.search-page`, `.about-page`, `.people`, `.role`, `.schedule`, `.subscribe`, `.newsletter`, `#comments`, `.replies`, `.reply-author`, `.reply-text`, `.reply-date`.
 - Produces: the complete stylesheet, served statically at `/styles.css` and shared by both rendering paths. `@view-transition { navigation: auto; }` is cross-document, so navigating between a router page and a static document animates.
 
@@ -2075,23 +2249,23 @@ EOF
 Append to `scripts/check-build.mjs`, immediately before the `console.log('\nSizes')` line:
 
 ```js
-console.log('\nTask 9: stylesheet')
-const cssSource = readFileSync('public/styles.css', 'utf8')
+console.log("\nTask 9: stylesheet");
+const cssSource = readFileSync("public/styles.css", "utf8");
 for (const needle of [
-  '@layer reset, tokens, layout, components, utilities;',
-  '@view-transition',
-  'light-dark(',
-  'oklch(',
-  '@container',
-  'anchor-name:',
-  'text-wrap: balance',
-  'text-wrap: pretty',
+  "@layer reset, tokens, layout, components, utilities;",
+  "@view-transition",
+  "light-dark(",
+  "oklch(",
+  "@container",
+  "anchor-name:",
+  "text-wrap: balance",
+  "text-wrap: pretty",
 ]) {
-  check(`styles.css uses ${needle}`, cssSource.includes(needle))
+  check(`styles.css uses ${needle}`, cssSource.includes(needle));
 }
-const lineCount = cssSource.split('\n').length
-check('styles.css is under 250 lines', lineCount < 250, `${lineCount} lines`)
-check('styles.css is served verbatim', file('styles.css') === cssSource)
+const lineCount = cssSource.split("\n").length;
+check("styles.css is under 250 lines", lineCount < 250, `${lineCount} lines`);
+check("styles.css is served verbatim", file("styles.css") === cssSource);
 ```
 
 - [ ] **Step 2: Run the check to verify it fails**
@@ -2105,12 +2279,33 @@ Expected: FAIL — everything except the `@layer` line and the line count.
 @layer reset, tokens, layout, components, utilities;
 
 @layer reset {
-  *, *::before, *::after { box-sizing: border-box; }
-  * { margin: 0; }
-  img, video, audio { display: block; max-inline-size: 100%; }
-  button, input { font: inherit; color: inherit; }
-  ul, ol { list-style: none; padding: 0; }
-  a { color: inherit; }
+  *,
+  *::before,
+  *::after {
+    box-sizing: border-box;
+  }
+  * {
+    margin: 0;
+  }
+  img,
+  video,
+  audio {
+    display: block;
+    max-inline-size: 100%;
+  }
+  button,
+  input {
+    font: inherit;
+    color: inherit;
+  }
+  ul,
+  ol {
+    list-style: none;
+    padding: 0;
+  }
+  a {
+    color: inherit;
+  }
 }
 
 @layer tokens {
@@ -2125,12 +2320,15 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     --measure: 68ch;
     --gap: 1.25rem;
     --radius: 0.5rem;
-    --font: ui-sans-serif, system-ui, -apple-system, "Segoe UI", Roboto, sans-serif;
+    --font:
+      ui-sans-serif, system-ui, -apple-system, "Segoe UI", Roboto, sans-serif;
     --mono: ui-monospace, "SF Mono", Menlo, monospace;
   }
 
   /* Cross-document: applies to static documents and router pages alike. */
-  @view-transition { navigation: auto; }
+  @view-transition {
+    navigation: auto;
+  }
 }
 
 @layer layout {
@@ -2142,7 +2340,9 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     padding-inline: var(--gap);
   }
 
-  .site-header, .site-footer, main {
+  .site-header,
+  .site-footer,
+  main {
     max-inline-size: 78rem;
     margin-inline: auto;
   }
@@ -2157,29 +2357,71 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     border-block-end: 1px solid var(--rule);
   }
 
-  .site-nav { display: flex; gap: var(--gap); }
-  .site-name { font-weight: 700; text-decoration: none; }
-  .site-footer { padding-block: 2rem; color: var(--muted); }
-  main { padding-block: 2rem; display: grid; gap: 2rem; }
+  .site-nav {
+    display: flex;
+    gap: var(--gap);
+  }
+  .site-name {
+    font-weight: 700;
+    text-decoration: none;
+  }
+  .site-footer {
+    padding-block: 2rem;
+    color: var(--muted);
+  }
+  main {
+    padding-block: 2rem;
+    display: grid;
+    gap: 2rem;
+  }
 
   /* Container query: outline beside the transcript when the page is wide enough. */
-  .episode-page-main { container: episode / inline-size; }
-  .episode { display: grid; gap: var(--gap); align-items: start; }
+  .episode-page-main {
+    container: episode / inline-size;
+  }
+  .episode {
+    display: grid;
+    gap: var(--gap);
+    align-items: start;
+  }
 
   @container episode (inline-size > 56rem) {
-    .episode { grid-template-columns: 18rem minmax(0, 1fr); }
-    .toc { position: sticky; inset-block-start: var(--gap); max-block-size: 80svh; overflow-y: auto; }
+    .episode {
+      grid-template-columns: 18rem minmax(0, 1fr);
+    }
+    .toc {
+      position: sticky;
+      inset-block-start: var(--gap);
+      max-block-size: 80svh;
+      overflow-y: auto;
+    }
   }
 }
 
 @layer components {
-  h1, h2, h3 { text-wrap: balance; line-height: 1.2; }
-  h1 { font-size: clamp(1.6rem, 1.1rem + 2vw, 2.4rem); }
-  h2 { font-size: 1.25rem; margin-block-start: 1.5em; }
-  p, li { text-wrap: pretty; }
+  h1,
+  h2,
+  h3 {
+    text-wrap: balance;
+    line-height: 1.2;
+  }
+  h1 {
+    font-size: clamp(1.6rem, 1.1rem + 2vw, 2.4rem);
+  }
+  h2 {
+    font-size: 1.25rem;
+    margin-block-start: 1.5em;
+  }
+  p,
+  li {
+    text-wrap: pretty;
+  }
 
   /* Anchor positioning: the player pins to the transcript column. */
-  .transcript { anchor-name: --transcript; max-inline-size: var(--measure); }
+  .transcript {
+    anchor-name: --transcript;
+    max-inline-size: var(--measure);
+  }
 
   .player {
     background: var(--surface);
@@ -2187,7 +2429,9 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     border-radius: var(--radius);
     padding: 0.5rem;
   }
-  .player audio { inline-size: 100%; }
+  .player audio {
+    inline-size: 100%;
+  }
 
   @supports (position-anchor: --transcript) {
     @container episode (inline-size > 56rem) {
@@ -2202,10 +2446,21 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     }
   }
 
-  .outline { display: grid; gap: 0.35rem; }
-  .outline ol { padding-inline-start: 1rem; margin-block-start: 0.35rem; }
-  .outline a { text-decoration-color: var(--rule); }
-  .outbound { color: var(--muted); text-decoration: none; }
+  .outline {
+    display: grid;
+    gap: 0.35rem;
+  }
+  .outline ol {
+    padding-inline-start: 1rem;
+    margin-block-start: 0.35rem;
+  }
+  .outline a {
+    text-decoration-color: var(--rule);
+  }
+  .outbound {
+    color: var(--muted);
+    text-decoration: none;
+  }
 
   .ts {
     font-family: var(--mono);
@@ -2216,17 +2471,46 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     padding: 0;
     cursor: pointer;
   }
-  button.ts:hover, button.ts:focus-visible { text-decoration: underline; }
+  button.ts:hover,
+  button.ts:focus-visible {
+    text-decoration: underline;
+  }
 
-  .transcript section { margin-block-end: 2rem; }
-  .transcript h2 { scroll-margin-block-start: var(--gap); }
-  .segment { margin-block-start: 0.85em; }
-  .speaker { color: var(--accent); }
+  .transcript section {
+    margin-block-end: 2rem;
+  }
+  .transcript h2 {
+    scroll-margin-block-start: var(--gap);
+  }
+  .segment {
+    margin-block-start: 0.85em;
+  }
+  .speaker {
+    color: var(--accent);
+  }
 
-  .archive ul, .link-group ul, .people ul, .subscribe ul { display: grid; gap: 0.9rem; }
-  .archive li { padding-block-end: 0.9rem; border-block-end: 1px solid var(--rule); }
-  .archive time, .from, .role, .count { color: var(--muted); font-size: 0.85em; }
-  .archive time { display: block; font-family: var(--mono); }
+  .archive ul,
+  .link-group ul,
+  .people ul,
+  .subscribe ul {
+    display: grid;
+    gap: 0.9rem;
+  }
+  .archive li {
+    padding-block-end: 0.9rem;
+    border-block-end: 1px solid var(--rule);
+  }
+  .archive time,
+  .from,
+  .role,
+  .count {
+    color: var(--muted);
+    font-size: 0.85em;
+  }
+  .archive time {
+    display: block;
+    font-family: var(--mono);
+  }
 
   #link-filter {
     inline-size: min(100%, 24rem);
@@ -2235,17 +2519,37 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     border-radius: var(--radius);
     background: var(--surface);
   }
-  .link-group { margin-block-start: 1.5rem; }
-  .link-row { display: flex; flex-wrap: wrap; gap: 0.5rem; align-items: baseline; }
-  .from { text-decoration: none; }
-  .from .ts { margin-inline-start: 0.4rem; }
+  .link-group {
+    margin-block-start: 1.5rem;
+  }
+  .link-row {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 0.5rem;
+    align-items: baseline;
+  }
+  .from {
+    text-decoration: none;
+  }
+  .from .ts {
+    margin-inline-start: 0.4rem;
+  }
 
-  .people li { display: flex; gap: 0.7rem; align-items: center; }
-  .people img { border-radius: 50%; }
+  .people li {
+    display: flex;
+    gap: 0.7rem;
+    align-items: center;
+  }
+  .people img {
+    border-radius: 50%;
+  }
 
   .newsletter form {
-    display: flex; flex-wrap: wrap; gap: 0.5rem;
-    align-items: center; margin-block-start: 0.7rem;
+    display: flex;
+    flex-wrap: wrap;
+    gap: 0.5rem;
+    align-items: center;
+    margin-block-start: 0.7rem;
   }
   .newsletter input {
     flex: 1 1 16rem;
@@ -2255,19 +2559,35 @@ Expected: FAIL — everything except the `@layer` line and the line count.
     background: var(--surface);
   }
   .newsletter button {
-    padding: 0.5rem 1rem; border: 0; border-radius: var(--radius);
-    background: var(--accent); color: var(--bg); cursor: pointer;
+    padding: 0.5rem 1rem;
+    border: 0;
+    border-radius: var(--radius);
+    background: var(--accent);
+    color: var(--bg);
+    cursor: pointer;
+  }
+  .newsletter label {
+    position: absolute;
+    clip-path: inset(50%);
   }
-  .newsletter label { position: absolute; clip-path: inset(50%); }
 }
 
 @layer utilities {
-  [hidden] { display: none !important; }
-  :focus-visible { outline: 2px solid var(--accent); outline-offset: 2px; }
+  [hidden] {
+    display: none !important;
+  }
+  :focus-visible {
+    outline: 2px solid var(--accent);
+    outline-offset: 2px;
+  }
 
   @media (prefers-reduced-motion: reduce) {
-    @view-transition { navigation: none; }
-    *, *::before, *::after {
+    @view-transition {
+      navigation: none;
+    }
+    *,
+    *::before,
+    *::after {
       animation-duration: 0.01ms !important;
       transition-duration: 0.01ms !important;
     }
@@ -2302,9 +2622,11 @@ EOF
 ## Task 10: Full build verification, size measurement, README
 
 **Files:**
+
 - Modify: `scripts/check-build.mjs`, `README.md` (create if absent)
 
 **Interfaces:**
+
 - Consumes: everything above.
 - Produces: a passing full-build verification, recorded size numbers, and documented env vars and publish flow.
 
@@ -2313,75 +2635,91 @@ EOF
 In `scripts/check-build.mjs`, replace the entire block from `console.log('\nSizes')` through the final `process.exit(...)` line with:
 
 ```js
-console.log('\nTask 10: prerender coverage')
-const allSlugs = slugs()
+console.log("\nTask 10: prerender coverage");
+const allSlugs = slugs();
 const expected = [
-  'index.html',
-  'styles.css',
-  'links/index.html',
-  'search/index.html',
-  'about/index.html',
-  'feed.xml',
-  ...allSlugs.flatMap((s) => [`episodes/${s}/index.html`, `episodes/${s}/chapters.json`]),
-]
-for (const path of expected) hasFile(path)
-check(`prerendered all ${expected.length} expected outputs`, expected.every((p) => file(p) !== null))
-
-console.log('\nTask 10: standard.site discovery')
-const wellKnown = '.well-known/site.standard.publication'
+  "index.html",
+  "styles.css",
+  "links/index.html",
+  "search/index.html",
+  "about/index.html",
+  "feed.xml",
+  ...allSlugs.flatMap((s) => [
+    `episodes/${s}/index.html`,
+    `episodes/${s}/chapters.json`,
+  ]),
+];
+for (const path of expected) hasFile(path);
+check(
+  `prerendered all ${expected.length} expected outputs`,
+  expected.every((p) => file(p) !== null),
+);
+
+console.log("\nTask 10: standard.site discovery");
+const wellKnown = ".well-known/site.standard.publication";
 if (process.env.VITE_ATPROTO_PUBLICATION_URI) {
-  hasFile(wellKnown)
+  hasFile(wellKnown);
   check(
     `${wellKnown} contains an at:// URI`,
-    (file(wellKnown) || '').trim().startsWith('at://'),
-    JSON.stringify((file(wellKnown) || '').trim()),
-  )
+    (file(wellKnown) || "").trim().startsWith("at://"),
+    JSON.stringify((file(wellKnown) || "").trim()),
+  );
 } else {
-  console.log(`  skip  ${wellKnown} (VITE_ATPROTO_PUBLICATION_URI unset)`)
+  console.log(`  skip  ${wellKnown} (VITE_ATPROTO_PUBLICATION_URI unset)`);
 }
 
-console.log('\nTask 10: no duplicated transcript')
+console.log("\nTask 10: no duplicated transcript");
 for (const slug of allSlugs) {
-  const html = file(`episodes/${slug}/index.html`)
-  if (html === null) continue
+  const html = file(`episodes/${slug}/index.html`);
+  if (html === null) continue;
   // The first transcript segment's text is the marker: it must appear exactly once.
-  const m = html.match(/<p class="segment">(?:<strong class="speaker">[^<]*<\/strong>)?([^<]{40,120})/)
+  const m = html.match(
+    /<p class="segment">(?:<strong class="speaker">[^<]*<\/strong>)?([^<]{40,120})/,
+  );
   if (!m) {
-    check(`episodes/${slug}: found a transcript segment to sample`, false)
-    continue
+    check(`episodes/${slug}: found a transcript segment to sample`, false);
+    continue;
   }
-  const marker = m[1]
-  const count = html.split(marker).length - 1
-  check(`episodes/${slug}: transcript marker appears exactly once`, count === 1, `appears ${count}x`)
+  const marker = m[1];
+  const count = html.split(marker).length - 1;
+  check(
+    `episodes/${slug}: transcript marker appears exactly once`,
+    count === 1,
+    `appears ${count}x`,
+  );
   check(
     `episodes/${slug}: references zero /assets/*.js`,
     !/\/assets\/[^"']*\.js/.test(html),
-  )
+  );
 }
 
-console.log('\nSizes')
-const assetsDir = join(dist, 'assets')
+console.log("\nSizes");
+const assetsDir = join(dist, "assets");
 const jsFiles = existsSync(assetsDir)
-  ? readdirSync(assetsDir).filter((f) => f.endsWith('.js'))
-  : []
-let rawTotal = 0
-let gzipTotal = 0
+  ? readdirSync(assetsDir).filter((f) => f.endsWith(".js"))
+  : [];
+let rawTotal = 0;
+let gzipTotal = 0;
 for (const f of jsFiles) {
-  const buf = readFileSync(join(assetsDir, f))
-  rawTotal += buf.length
-  gzipTotal += gzipSync(buf).length
+  const buf = readFileSync(join(assetsDir, f));
+  rawTotal += buf.length;
+  gzipTotal += gzipSync(buf).length;
 }
-console.log(`  router-page JS (${jsFiles.length} files): ${rawTotal} B raw / ${gzipTotal} B gzip`)
-const sampleSlug = allSlugs.at(-1)
+console.log(
+  `  router-page JS (${jsFiles.length} files): ${rawTotal} B raw / ${gzipTotal} B gzip`,
+);
+const sampleSlug = allSlugs.at(-1);
 if (sampleSlug) {
-  const html = readFileSync(join(dist, 'episodes', sampleSlug, 'index.html'))
+  const html = readFileSync(join(dist, "episodes", sampleSlug, "index.html"));
   console.log(
     `  episode document ${sampleSlug}/index.html: ${html.length} B raw / ${gzipSync(html).length} B gzip, 0 external scripts`,
-  )
+  );
 }
-console.log('  spike baselines: router runtime 342267 B raw / 108613 B gzip; 100 KB fixture document 101837 B HTML')
+console.log(
+  "  spike baselines: router runtime 342267 B raw / 108613 B gzip; 100 KB fixture document 101837 B HTML",
+);
 
-process.exit(failures === 0 ? 0 : 1)
+process.exit(failures === 0 ? 0 : 1);
 ```
 
 - [ ] **Step 2: Run the full production build**
@@ -2402,11 +2740,13 @@ Expected: the two numbers are equal.
 - [ ] **Step 4a: Confirm the standard.site discovery file is emitted**
 
 Run:
+
 ```bash
 VITE_ATPROTO_PUBLICATION_URI=at://did:plc:example/site.standard.publication/self npm run build \
   && VITE_ATPROTO_PUBLICATION_URI=at://did:plc:example/site.standard.publication/self node scripts/check-build.mjs \
   && cat dist/client/.well-known/site.standard.publication
 ```
+
 Expected: the two `.well-known` lines `ok`, and the file's contents are the `at://` URI. Then `rm -rf public/.well-known` and rebuild without the variable so the working tree is clean; the check prints the `skip` line instead.
 
 - [ ] **Step 5: Confirm the crawl catches a broken internal link**
@@ -2417,7 +2757,7 @@ Temporarily add `<a href="/does-not-exist">x</a>` to the nav in `src/components/
 
 Append to `README.md` (create the file with this content if it does not exist):
 
-```markdown
+````markdown
 ## Site
 
 TanStack Start, fully prerendered to static files. There is no server runtime in
@@ -2433,19 +2773,19 @@ Two rendering paths share one `Document` component and one stylesheet:
   the framework runtime, about 109 KB gzipped. RSC is enabled in the build for
   future supporting pages.
 
-Transcripts are deliberately *not* rendered through RSC or a router loader:
+Transcripts are deliberately _not_ rendered through RSC or a router loader:
 both inline a second serialized copy of the content for hydration, which
 doubles an hour-long episode page.
 
 ### Commands
 
-| Command | Effect |
-| --- | --- |
-| `npm run dev` | dev server on http://localhost:3000 |
-| `npm run build` | prerenders every route to `dist/client` |
+| Command                           | Effect                                                      |
+| --------------------------------- | ----------------------------------------------------------- |
+| `npm run dev`                     | dev server on http://localhost:3000                         |
+| `npm run build`                   | prerenders every route to `dist/client`                     |
 | `npx pagefind --site dist/client` | builds the search index (Netlify runs this after the build) |
-| `npm run check` | asserts the build output and prints sizes |
-| `npm test` | unit tests |
+| `npm run check`                   | asserts the build output and prints sizes                   |
+| `npm test`                        | unit tests                                                  |
 
 ### Environment variables
 
@@ -2455,6 +2795,7 @@ lives in one gitignored `.env` at the repo root. Set it up with:
 ```bash
 cp .env.example .env
 ```
+````
 
 `.env.example` is committed and is the authoritative list, one comment per
 variable; this README does not duplicate it. Vite reads `.env` natively for the
@@ -2507,7 +2848,8 @@ page reads fine without it.
 
 Every one is a static file after the build. A broken internal link fails the
 build, because the prerenderer crawls `<a href>` from `/`.
-```
+
+````
 
 - [ ] **Step 7: Commit, pasting the recorded sizes into the message**
 
@@ -2527,7 +2869,7 @@ Spike baselines: router runtime 342267 B raw / 108613 B gzip;
 Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 EOF
 )"
-```
+````
 
 - [ ] **Step 8: Confirm nothing is left uncommitted that should be**
 
@@ -2540,37 +2882,37 @@ Expected: only untracked files belonging to other work (audio, SRT, `dist/`, `no
 
 **1. Spec coverage — the "Site" section, requirement by requirement:**
 
-| Spec requirement | Task |
-| --- | --- |
-| TanStack Start, React 19, Vite; RSC enabled through the Vite RSC plugin | 1 |
-| Full static prerender with link crawling; broken internal link fails the build | 1, 4, 10 (Step 5 verifies it) |
-| Netlify via the official plugin | 1 |
-| Pagefind run after build | 1 (`netlify.toml`), 7 |
-| Site name and canonical URL from environment | 1, 10 |
-| **Rendering split:** static documents for `/episodes/<slug>` and `/links`, `renderToStaticMarkup`, content once, no hydration, small inline script | 3, 5 |
-| **Rendering split:** router pages for `/`, `/about`, `/search`; RSC stays enabled; home player is a client component | 1, 4, 7, 8 |
-| Both paths share one `Document` for head, stylesheet and navigation | 1 |
-| "Why not RSC or loaders for transcripts" — no duplicated copy | Global Constraints; asserted in 10 |
-| `/` — latest episode with player and outline, then the archive | 4 |
-| `/episodes/<yyyy-mm>` — player, outline as TOC, transcript | 3 |
-| `/links` — grouped and filterable | 5 |
-| `/search` — Pagefind UI | 7 |
-| `/about` — hosts, guests, subscribe links, live schedule, newsletter form | 8 |
-| `/feed.xml` — item body is the rendered outline | 6 |
-| `/episodes/<yyyy-mm>/chapters.json` — Podcasting 2.0 | 6 |
-| Link index: text, URL, hostname, slug, title, timestamp; grouped by host sorted by count; items without URLs excluded | 2, 5 |
-| Newsletter: RSS feed + Buttondown embed form, username from environment | 6, 8 |
-| CSS: five cascade layers, no preprocessor/framework/CSS-in-JS | 9, Global Constraints |
-| Container queries on the episode layout | 9 |
-| **Cross-document** view transitions | 9 |
-| `light-dark()` with oklch | 9 |
-| Anchor positioning for the player | 9 |
-| `text-wrap: balance` / `pretty` | 9 |
-| `/.well-known/site.standard.publication` from `VITE_ATPROTO_PUBLICATION_URI` | 1 (inline Vite plugin), asserted in 10 |
-| `<link rel="site.standard.document">` on episode pages when `atUri` is set | 3 |
-| "Reply on Bluesky" link and a comments section fetching the thread with a small inline script; page reads fine without it | 3 (`COMMENTS_SCRIPT`, `bskyPostToAtUri`) |
-| One `.env` for build variables and script credentials; `.env.example` committed | Global Constraints, 10 (README); the file itself is created in the content pipeline plan's Task 7 |
-| Testing: the prerender crawl is the integration test; no browser E2E | 10 |
+| Spec requirement                                                                                                                                   | Task                                                                                              |
+| -------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------- |
+| TanStack Start, React 19, Vite; RSC enabled through the Vite RSC plugin                                                                            | 1                                                                                                 |
+| Full static prerender with link crawling; broken internal link fails the build                                                                     | 1, 4, 10 (Step 5 verifies it)                                                                     |
+| Netlify via the official plugin                                                                                                                    | 1                                                                                                 |
+| Pagefind run after build                                                                                                                           | 1 (`netlify.toml`), 7                                                                             |
+| Site name and canonical URL from environment                                                                                                       | 1, 10                                                                                             |
+| **Rendering split:** static documents for `/episodes/<slug>` and `/links`, `renderToStaticMarkup`, content once, no hydration, small inline script | 3, 5                                                                                              |
+| **Rendering split:** router pages for `/`, `/about`, `/search`; RSC stays enabled; home player is a client component                               | 1, 4, 7, 8                                                                                        |
+| Both paths share one `Document` for head, stylesheet and navigation                                                                                | 1                                                                                                 |
+| "Why not RSC or loaders for transcripts" — no duplicated copy                                                                                      | Global Constraints; asserted in 10                                                                |
+| `/` — latest episode with player and outline, then the archive                                                                                     | 4                                                                                                 |
+| `/episodes/<yyyy-mm>` — player, outline as TOC, transcript                                                                                         | 3                                                                                                 |
+| `/links` — grouped and filterable                                                                                                                  | 5                                                                                                 |
+| `/search` — Pagefind UI                                                                                                                            | 7                                                                                                 |
+| `/about` — hosts, guests, subscribe links, live schedule, newsletter form                                                                          | 8                                                                                                 |
+| `/feed.xml` — item body is the rendered outline                                                                                                    | 6                                                                                                 |
+| `/episodes/<yyyy-mm>/chapters.json` — Podcasting 2.0                                                                                               | 6                                                                                                 |
+| Link index: text, URL, hostname, slug, title, timestamp; grouped by host sorted by count; items without URLs excluded                              | 2, 5                                                                                              |
+| Newsletter: RSS feed + Buttondown embed form, username from environment                                                                            | 6, 8                                                                                              |
+| CSS: five cascade layers, no preprocessor/framework/CSS-in-JS                                                                                      | 9, Global Constraints                                                                             |
+| Container queries on the episode layout                                                                                                            | 9                                                                                                 |
+| **Cross-document** view transitions                                                                                                                | 9                                                                                                 |
+| `light-dark()` with oklch                                                                                                                          | 9                                                                                                 |
+| Anchor positioning for the player                                                                                                                  | 9                                                                                                 |
+| `text-wrap: balance` / `pretty`                                                                                                                    | 9                                                                                                 |
+| `/.well-known/site.standard.publication` from `VITE_ATPROTO_PUBLICATION_URI`                                                                       | 1 (inline Vite plugin), asserted in 10                                                            |
+| `<link rel="site.standard.document">` on episode pages when `atUri` is set                                                                         | 3                                                                                                 |
+| "Reply on Bluesky" link and a comments section fetching the thread with a small inline script; page reads fine without it                          | 3 (`COMMENTS_SCRIPT`, `bskyPostToAtUri`)                                                          |
+| One `.env` for build variables and script credentials; `.env.example` committed                                                                    | Global Constraints, 10 (README); the file itself is created in the content pipeline plan's Task 7 |
+| Testing: the prerender crawl is the integration test; no browser E2E                                                                               | 10                                                                                                |
 
 One spec item is deliberately **not** a task: "verification against the
 standard.site validator is part of the final build check." The validator is a
@@ -2583,6 +2925,7 @@ listed in the spec's "Open items outside this repository".
 **2. Placeholder scan:** every code step contains complete, runnable code. Two named soft spots, each with a concrete action rather than a deferral: the Spotify and Apple show IDs (Task 8, flagged for verification against the live podcast page) and the `.tsx`-import fallback in Task 6 Step 7. Task 2 Step 1 verifies the parser's `time` type and states that `toSeconds` already absorbs either shape. No "TBD", no "handle edge cases", no "similar to Task N".
 
 **3. Type consistency:**
+
 - `hms` / `toSeconds` live in `src/content/time.ts` and are imported from there by `EpisodeBody.tsx`, `links.ts`, `feed.ts` and `links.tsx` — never from `load.ts`.
 - `loadEpisodes` / `getEpisode` (`load.ts`) are used with those exact names in Tasks 3, 4, 5, 6, 8, always via dynamic import inside a handler.
 - `buildLinkIndex` / `groupByHost` / `LinkEntry` (`links.ts`) used with those names in Task 5.
@@ -2594,6 +2937,7 @@ listed in the spec's "Open items outside this repository".
 - `SITE_NAME` / `SITE_URL` are exported from `src/components/Document.tsx` only — the earlier draft exported them from `__root.tsx`, which the static-document path cannot import.
 
 **Five fixes applied inline during review:**
+
 1. `hms`/`toSeconds` were originally in `load.ts`; `EpisodeBody.tsx` imports them and is itself imported by the router route `index.tsx`, which would have pulled `node:fs/promises` into the client bundle. Split into `src/content/time.ts` (Task 2).
 2. `SITE_NAME`/`SITE_URL` moved from `__root.tsx` to `Document.tsx` so both rendering paths can import them.
 3. The container-query selector is `.episode-page-main`, not `.episode-page`, because `Document` puts `bodyClass` on `<body>` and `` `${bodyClass}-main` `` on `<main>`. Task 9 matches.
diff --git a/docs/superpowers/specs/2026-09-08-tmir-site-design.md b/docs/superpowers/specs/2026-09-08-tmir-site-design.md
index 5369c70..570de6a 100644
--- a/docs/superpowers/specs/2026-09-08-tmir-site-design.md
+++ b/docs/superpowers/specs/2026-09-08-tmir-site-design.md
@@ -86,26 +86,26 @@ file per episode at `content/episodes/<yyyy>-<mm>.md`.
 
 Hand-written fields:
 
-| Field         | Type   | Notes                                                   |
-| ------------- | ------ | ------------------------------------------------------- |
-| `title`       | string | Episode title as published                              |
-| `date`        | date   | Recording date, ISO                                     |
-| `description` | string | One paragraph, used for meta tags and the archive list  |
-| `time`        | string | Live show time, e.g. `2pm PT / 9pm GMT`, optional       |
-| `location`    | string | Live show location, optional                            |
+| Field         | Type   | Notes                                                  |
+| ------------- | ------ | ------------------------------------------------------ |
+| `title`       | string | Episode title as published                             |
+| `date`        | date   | Recording date, ISO                                    |
+| `description` | string | One paragraph, used for meta tags and the archive list |
+| `time`        | string | Live show time, e.g. `2pm PT / 9pm GMT`, optional      |
+| `location`    | string | Live show location, optional                           |
 
 Fields written by the ingest script, never by hand:
 
-| Field           | Type     | Source                                    |
-| --------------- | -------- | ----------------------------------------- |
-| `transistorId`  | string   | Share ID from the episode `link`          |
-| `audioUrl`      | string   | Enclosure URL                             |
-| `duration`      | number   | Seconds, from `itunes:duration`           |
-| `season`        | number   | `podcast:season`                          |
-| `episode`       | number   | `podcast:episode`                         |
-| `people`        | list     | `podcast:person`: name, role, href, img   |
-| `bskyPostUrl`   | string   | "Reply on Bluesky" link in the description |
-| `atUri`         | string   | Written by `publish-atproto`, not ingest   |
+| Field          | Type   | Source                                     |
+| -------------- | ------ | ------------------------------------------ |
+| `transistorId` | string | Share ID from the episode `link`           |
+| `audioUrl`     | string | Enclosure URL                              |
+| `duration`     | number | Seconds, from `itunes:duration`            |
+| `season`       | number | `podcast:season`                           |
+| `episode`      | number | `podcast:episode`                          |
+| `people`       | list   | `podcast:person`: name, role, href, img    |
+| `bskyPostUrl`  | string | "Reply on Bluesky" link in the description |
+| `atUri`        | string | Written by `publish-atproto`, not ingest   |
 
 Ingest matches feed items to files by the `yyyy-mm` in the feed title,
 which every TMiR title carries as `TMiR yyyy-mm:` or as a month name plus
@@ -211,15 +211,15 @@ transcript can still be pushed.
 
 ### Routes
 
-| Route                 | Purpose                                                      |
-| --------------------- | ------------------------------------------------------------ |
-| `/`                   | Latest episode with player and outline, then the archive     |
-| `/episodes/<yyyy-mm>` | Player, outline as table of contents, transcript             |
-| `/links`              | Every outline link across all episodes, grouped and filterable |
-| `/search`             | Pagefind UI                                                  |
-| `/about`              | Hosts, guests, subscribe links, live schedule, newsletter form |
-| `/feed.xml`           | RSS of episode pages; item body is the rendered outline      |
-| `/episodes/<yyyy-mm>/chapters.json` | Podcasting 2.0 chapters derived from the outline |
+| Route                               | Purpose                                                        |
+| ----------------------------------- | -------------------------------------------------------------- |
+| `/`                                 | Latest episode with player and outline, then the archive       |
+| `/episodes/<yyyy-mm>`               | Player, outline as table of contents, transcript               |
+| `/links`                            | Every outline link across all episodes, grouped and filterable |
+| `/search`                           | Pagefind UI                                                    |
+| `/about`                            | Hosts, guests, subscribe links, live schedule, newsletter form |
+| `/feed.xml`                         | RSS of episode pages; item body is the rendered outline        |
+| `/episodes/<yyyy-mm>/chapters.json` | Podcasting 2.0 chapters derived from the outline               |
 
 ### Rendering split
 
diff --git a/package-lock.json b/package-lock.json
index 9319899..0799764 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -15,13 +15,16 @@
       },
       "devDependencies": {
         "@netlify/vite-plugin-tanstack-start": "1.3.19",
+        "@resvg/resvg-js": "^2.6.2",
         "@types/node": "22.12.0",
         "@types/react": "19.2.0",
         "@types/react-dom": "19.2.0",
         "@vitejs/plugin-react": "6.1.1",
         "@vitejs/plugin-rsc": "0.5.34",
+        "jpeg-js": "^0.4.4",
         "pagefind": "1.5.2",
         "prettier": "^3.3.0",
+        "satori": "^0.33.4",
         "typescript": "6.0.3",
         "vite": "8.2.2"
       }
@@ -2629,6 +2632,246 @@
       "inBundle": true,
       "license": "MIT"
     },
+    "node_modules/@resvg/resvg-js": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js/-/resvg-js-2.6.2.tgz",
+      "integrity": "sha512-xBaJish5OeGmniDj9cW5PRa/PtmuVU3ziqrbr5xJj901ZDN4TosrVaNZpEiLZAxdfnhAe7uQ7QFWfjPe9d9K2Q==",
+      "dev": true,
+      "license": "MPL-2.0",
+      "engines": {
+        "node": ">= 10"
+      },
+      "optionalDependencies": {
+        "@resvg/resvg-js-android-arm-eabi": "2.6.2",
+        "@resvg/resvg-js-android-arm64": "2.6.2",
+        "@resvg/resvg-js-darwin-arm64": "2.6.2",
+        "@resvg/resvg-js-darwin-x64": "2.6.2",
+        "@resvg/resvg-js-linux-arm-gnueabihf": "2.6.2",
+        "@resvg/resvg-js-linux-arm64-gnu": "2.6.2",
+        "@resvg/resvg-js-linux-arm64-musl": "2.6.2",
+        "@resvg/resvg-js-linux-x64-gnu": "2.6.2",
+        "@resvg/resvg-js-linux-x64-musl": "2.6.2",
+        "@resvg/resvg-js-win32-arm64-msvc": "2.6.2",
+        "@resvg/resvg-js-win32-ia32-msvc": "2.6.2",
+        "@resvg/resvg-js-win32-x64-msvc": "2.6.2"
+      }
+    },
+    "node_modules/@resvg/resvg-js-android-arm-eabi": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-android-arm-eabi/-/resvg-js-android-arm-eabi-2.6.2.tgz",
+      "integrity": "sha512-FrJibrAk6v29eabIPgcTUMPXiEz8ssrAk7TXxsiZzww9UTQ1Z5KAbFJs+Z0Ez+VZTYgnE5IQJqBcoSiMebtPHA==",
+      "cpu": [
+        "arm"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "android"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-android-arm64": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-android-arm64/-/resvg-js-android-arm64-2.6.2.tgz",
+      "integrity": "sha512-VcOKezEhm2VqzXpcIJoITuvUS/fcjIw5NA/w3tjzWyzmvoCdd+QXIqy3FBGulWdClvp4g+IfUemigrkLThSjAQ==",
+      "cpu": [
+        "arm64"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "android"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-darwin-arm64": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-darwin-arm64/-/resvg-js-darwin-arm64-2.6.2.tgz",
+      "integrity": "sha512-nmok2LnAd6nLUKI16aEB9ydMC6Lidiiq2m1nEBDR1LaaP7FGs4AJ90qDraxX+CWlVuRlvNjyYJTNv8qFjtL9+A==",
+      "cpu": [
+        "arm64"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "darwin"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-darwin-x64": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-darwin-x64/-/resvg-js-darwin-x64-2.6.2.tgz",
+      "integrity": "sha512-GInyZLjgWDfsVT6+SHxQVRwNzV0AuA1uqGsOAW+0th56J7Nh6bHHKXHBWzUrihxMetcFDmQMAX1tZ1fZDYSRsw==",
+      "cpu": [
+        "x64"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "darwin"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-linux-arm-gnueabihf": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-linux-arm-gnueabihf/-/resvg-js-linux-arm-gnueabihf-2.6.2.tgz",
+      "integrity": "sha512-YIV3u/R9zJbpqTTNwTZM5/ocWetDKGsro0SWp70eGEM9eV2MerWyBRZnQIgzU3YBnSBQ1RcxRZvY/UxwESfZIw==",
+      "cpu": [
+        "arm"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "linux"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-linux-arm64-gnu": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-linux-arm64-gnu/-/resvg-js-linux-arm64-gnu-2.6.2.tgz",
+      "integrity": "sha512-zc2BlJSim7YR4FZDQ8OUoJg5holYzdiYMeobb9pJuGDidGL9KZUv7SbiD4E8oZogtYY42UZEap7dqkkYuA91pg==",
+      "cpu": [
+        "arm64"
+      ],
+      "dev": true,
+      "libc": [
+        "glibc"
+      ],
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "linux"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-linux-arm64-musl": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-linux-arm64-musl/-/resvg-js-linux-arm64-musl-2.6.2.tgz",
+      "integrity": "sha512-3h3dLPWNgSsD4lQBJPb4f+kvdOSJHa5PjTYVsWHxLUzH4IFTJUAnmuWpw4KqyQ3NA5QCyhw4TWgxk3jRkQxEKg==",
+      "cpu": [
+        "arm64"
+      ],
+      "dev": true,
+      "libc": [
+        "musl"
+      ],
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "linux"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-linux-x64-gnu": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-linux-x64-gnu/-/resvg-js-linux-x64-gnu-2.6.2.tgz",
+      "integrity": "sha512-IVUe+ckIerA7xMZ50duAZzwf1U7khQe2E0QpUxu5MBJNao5RqC0zwV/Zm965vw6D3gGFUl7j4m+oJjubBVoftw==",
+      "cpu": [
+        "x64"
+      ],
+      "dev": true,
+      "libc": [
+        "glibc"
+      ],
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "linux"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-linux-x64-musl": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-linux-x64-musl/-/resvg-js-linux-x64-musl-2.6.2.tgz",
+      "integrity": "sha512-UOf83vqTzoYQO9SZ0fPl2ZIFtNIz/Rr/y+7X8XRX1ZnBYsQ/tTb+cj9TE+KHOdmlTFBxhYzVkP2lRByCzqi4jQ==",
+      "cpu": [
+        "x64"
+      ],
+      "dev": true,
+      "libc": [
+        "musl"
+      ],
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "linux"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-win32-arm64-msvc": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-win32-arm64-msvc/-/resvg-js-win32-arm64-msvc-2.6.2.tgz",
+      "integrity": "sha512-7C/RSgCa+7vqZ7qAbItfiaAWhyRSoD4l4BQAbVDqRRsRgY+S+hgS3in0Rxr7IorKUpGE69X48q6/nOAuTJQxeQ==",
+      "cpu": [
+        "arm64"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "win32"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-win32-ia32-msvc": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-win32-ia32-msvc/-/resvg-js-win32-ia32-msvc-2.6.2.tgz",
+      "integrity": "sha512-har4aPAlvjnLcil40AC77YDIk6loMawuJwFINEM7n0pZviwMkMvjb2W5ZirsNOZY4aDbo5tLx0wNMREp5Brk+w==",
+      "cpu": [
+        "ia32"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "win32"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
+    "node_modules/@resvg/resvg-js-win32-x64-msvc": {
+      "version": "2.6.2",
+      "resolved": "https://registry.npmjs.org/@resvg/resvg-js-win32-x64-msvc/-/resvg-js-win32-x64-msvc-2.6.2.tgz",
+      "integrity": "sha512-ZXtYhtUr5SSaBrUDq7DiyjOFJqBVL/dOBN7N/qmi/pO0IgiWW/f/ue3nbvu9joWE5aAKDoIzy/CxsY0suwGosQ==",
+      "cpu": [
+        "x64"
+      ],
+      "dev": true,
+      "license": "MPL-2.0",
+      "optional": true,
+      "os": [
+        "win32"
+      ],
+      "engines": {
+        "node": ">= 10"
+      }
+    },
     "node_modules/@rolldown/binding-android-arm-eabi": {
       "version": "1.2.7",
       "resolved": "https://registry.npmjs.org/@rolldown/binding-android-arm-eabi/-/binding-android-arm-eabi-1.2.7.tgz",
@@ -2917,6 +3160,23 @@
         }
       }
     },
+    "node_modules/@shuding/opentype.js": {
+      "version": "1.4.0-beta.0",
+      "resolved": "https://registry.npmjs.org/@shuding/opentype.js/-/opentype.js-1.4.0-beta.0.tgz",
+      "integrity": "sha512-3NgmNyH3l/Hv6EvsWJbsvpcpUba6R8IREQ83nH83cyakCw7uM1arZKNfHwv1Wz6jgqrF/j4x5ELvR6PnK9nTcA==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "fflate": "^0.7.3",
+        "string.prototype.codepointat": "^0.2.1"
+      },
+      "bin": {
+        "ot": "bin/ot"
+      },
+      "engines": {
+        "node": ">= 8.0.0"
+      }
+    },
     "node_modules/@so-ric/colorspace": {
       "version": "1.1.6",
       "resolved": "https://registry.npmjs.org/@so-ric/colorspace/-/colorspace-1.1.6.tgz",
@@ -4499,6 +4759,16 @@
         "node": "*"
       }
     },
+    "node_modules/camelize": {
+      "version": "1.0.1",
+      "resolved": "https://registry.npmjs.org/camelize/-/camelize-1.0.1.tgz",
+      "integrity": "sha512-dU+Tx2fsypxTgtLoE36npi3UqcjSSMNYfkqgmoEhtZrraP5VWq0K7FkWVTYa8eMPtnU/G2txVsfdCJTn9uzpuQ==",
+      "dev": true,
+      "license": "MIT",
+      "funding": {
+        "url": "https://github.com/sponsors/ljharb"
+      }
+    },
     "node_modules/caniuse-lite": {
       "version": "1.0.30001810",
       "resolved": "https://registry.npmjs.org/caniuse-lite/-/caniuse-lite-1.0.30001810.tgz",
@@ -4821,6 +5091,40 @@
         }
       }
     },
+    "node_modules/css-background-parser": {
+      "version": "0.1.0",
+      "resolved": "https://registry.npmjs.org/css-background-parser/-/css-background-parser-0.1.0.tgz",
+      "integrity": "sha512-2EZLisiZQ+7m4wwur/qiYJRniHX4K5Tc9w93MT3AS0WS1u5kaZ4FKXlOTBhOjc+CgEgPiGY+fX1yWD8UwpEqUA==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/css-box-shadow": {
+      "version": "1.0.0-3",
+      "resolved": "https://registry.npmjs.org/css-box-shadow/-/css-box-shadow-1.0.0-3.tgz",
+      "integrity": "sha512-9jaqR6e7Ohds+aWwmhe6wILJ99xYQbfmK9QQB9CcMjDbTxPZjwEmUQpU91OG05Xgm8BahT5fW+svbsQGjS/zPg==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/css-color-keywords": {
+      "version": "1.0.0",
+      "resolved": "https://registry.npmjs.org/css-color-keywords/-/css-color-keywords-1.0.0.tgz",
+      "integrity": "sha512-FyyrDHZKEjXDpNJYvVsV960FiqQyXc/LlYmsxl2BcdMb2WPx0OGRVgTg55rPSyLSNMqP52R9r8geSp7apN3Ofg==",
+      "dev": true,
+      "license": "ISC",
+      "engines": {
+        "node": ">=4"
+      }
+    },
+    "node_modules/css-gradient-parser": {
+      "version": "0.0.17",
+      "resolved": "https://registry.npmjs.org/css-gradient-parser/-/css-gradient-parser-0.0.17.tgz",
+      "integrity": "sha512-w2Xy9UMMwlKtou0vlRnXvWglPAceXCTtcmVSo8ZBUvqCV5aXEFP/PC6d+I464810I9FT++UACwTD5511bmGPUg==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=16"
+      }
+    },
     "node_modules/css-select": {
       "version": "6.0.0",
       "resolved": "https://registry.npmjs.org/css-select/-/css-select-6.0.0.tgz",
@@ -4838,6 +5142,18 @@
         "url": "https://github.com/sponsors/fb55"
       }
     },
+    "node_modules/css-to-react-native": {
+      "version": "3.2.0",
+      "resolved": "https://registry.npmjs.org/css-to-react-native/-/css-to-react-native-3.2.0.tgz",
+      "integrity": "sha512-e8RKaLXMOFii+02mOlqwjbD00KSEKqblnpO9e++1aXS1fPQOpS1YoqdVHBqPjHNoxeF2mimzVqawm2KCbEdtHQ==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "camelize": "^1.0.0",
+        "css-color-keywords": "^1.0.0",
+        "postcss-value-parser": "^4.0.2"
+      }
+    },
     "node_modules/css-tree": {
       "version": "3.2.1",
       "resolved": "https://registry.npmjs.org/css-tree/-/css-tree-3.2.1.tgz",
@@ -5386,6 +5702,16 @@
       "dev": true,
       "license": "MIT"
     },
+    "node_modules/emoji-regex-xs": {
+      "version": "2.0.1",
+      "resolved": "https://registry.npmjs.org/emoji-regex-xs/-/emoji-regex-xs-2.0.1.tgz",
+      "integrity": "sha512-1QFuh8l7LqUcKe24LsPUNzjrzJQ7pgRwp1QMcZ5MX6mFplk2zQ08NVCM84++1cveaUUYtcCYHmeFEuNg16sU4g==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=10.0.0"
+      }
+    },
     "node_modules/empathic": {
       "version": "2.0.1",
       "resolved": "https://registry.npmjs.org/empathic/-/empathic-2.0.1.tgz",
@@ -5658,6 +5984,13 @@
         "node": ">=6"
       }
     },
+    "node_modules/escape-html": {
+      "version": "1.0.3",
+      "resolved": "https://registry.npmjs.org/escape-html/-/escape-html-1.0.3.tgz",
+      "integrity": "sha512-NiSupZ4OeuGwr68lGIeym/ksIZMJodUGOSCZ/FSnTxcrekbvqrgdUxlJOMpijaKZVjAJrWrGs/6Jy8OMuyj9ow==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/escape-string-regexp": {
       "version": "5.0.0",
       "resolved": "https://registry.npmjs.org/escape-string-regexp/-/escape-string-regexp-5.0.0.tgz",
@@ -5960,6 +6293,13 @@
       "integrity": "sha512-YoZjBdafyLIop9lSxXVI33oLD5kN31q4Td+CasofLLYeLXRFeOsuOw0Uo+XNRi9PZlbfdlN2GmRtm4tCEQ9/KA==",
       "license": "MIT"
     },
+    "node_modules/fflate": {
+      "version": "0.7.3",
+      "resolved": "https://registry.npmjs.org/fflate/-/fflate-0.7.3.tgz",
+      "integrity": "sha512-0Zz1jOzJWERhyhsimS54VTqOteCNwRtIlh8isdL0AXLo0g7xNTfTL7oWrkmCnPhZGocKIkWHBistBrrpoNH3aw==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/figures": {
       "version": "6.1.0",
       "resolved": "https://registry.npmjs.org/figures/-/figures-6.1.0.tgz",
@@ -6411,6 +6751,13 @@
         "uncrypto": "^0.1.3"
       }
     },
+    "node_modules/harfbuzzjs": {
+      "version": "0.10.0",
+      "resolved": "https://registry.npmjs.org/harfbuzzjs/-/harfbuzzjs-0.10.0.tgz",
+      "integrity": "sha512-SN8LVwCOzvTq3OPNd0+EAghgXugItz56wst567D1vs7LnnKGWprhi3EG58aVOBwhN8HEbQxL4I9NDWkcXUutRw==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/has-bigints": {
       "version": "1.1.0",
       "resolved": "https://registry.npmjs.org/has-bigints/-/has-bigints-1.1.0.tgz",
@@ -6505,6 +6852,19 @@
         "node": ">= 0.4"
       }
     },
+    "node_modules/hex-rgb": {
+      "version": "4.3.0",
+      "resolved": "https://registry.npmjs.org/hex-rgb/-/hex-rgb-4.3.0.tgz",
+      "integrity": "sha512-Ox1pJVrDCyGHMG9CFg1tmrRUMRPRsAWYc/PinY0XzJU4K7y7vjNoLKIQ7BR5UJMCxNN8EM1MNDmHWA/B3aZUuw==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=6"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/sindresorhus"
+      }
+    },
     "node_modules/hosted-git-info": {
       "version": "7.0.2",
       "resolved": "https://registry.npmjs.org/hosted-git-info/-/hosted-git-info-7.0.2.tgz",
@@ -7337,6 +7697,13 @@
         "jiti": "lib/jiti-cli.mjs"
       }
     },
+    "node_modules/jpeg-js": {
+      "version": "0.4.4",
+      "resolved": "https://registry.npmjs.org/jpeg-js/-/jpeg-js-0.4.4.tgz",
+      "integrity": "sha512-WZzeDOEtTOBK4Mdsar0IqEU5sMr3vSV2RqkAIzUEV2BHnUfKGyswWFPFwK5EeDo93K3FohSHbLAjj0s1Wzd+dg==",
+      "dev": true,
+      "license": "BSD-3-Clause"
+    },
     "node_modules/js-tokens": {
       "version": "4.0.0",
       "resolved": "https://registry.npmjs.org/js-tokens/-/js-tokens-4.0.0.tgz",
@@ -7824,6 +8191,27 @@
         "url": "https://opencollective.com/parcel"
       }
     },
+    "node_modules/linebreak": {
+      "version": "1.1.0",
+      "resolved": "https://registry.npmjs.org/linebreak/-/linebreak-1.1.0.tgz",
+      "integrity": "sha512-MHp03UImeVhB7XZtjd0E4n6+3xr5Dq/9xI/5FptGk5FrbDR3zagPa2DS6U8ks/3HjbKWG9Q1M2ufOzxV2qLYSQ==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "base64-js": "0.0.8",
+        "unicode-trie": "^2.0.0"
+      }
+    },
+    "node_modules/linebreak/node_modules/base64-js": {
+      "version": "0.0.8",
+      "resolved": "https://registry.npmjs.org/base64-js/-/base64-js-0.0.8.tgz",
+      "integrity": "sha512-3XSA2cR/h/73EzlXXdU6YNycmYI7+kicTxks4eJg2g39biHR84slg2+des+p7iHYhbRg/udIS4TD53WabcOUkw==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">= 0.4"
+      }
+    },
     "node_modules/listhen": {
       "version": "1.10.1",
       "resolved": "https://registry.npmjs.org/listhen/-/listhen-1.10.1.tgz",
@@ -8689,6 +9077,24 @@
         "@pagefind/windows-x64": "1.5.2"
       }
     },
+    "node_modules/pako": {
+      "version": "0.2.9",
+      "resolved": "https://registry.npmjs.org/pako/-/pako-0.2.9.tgz",
+      "integrity": "sha512-NUcwaKxUxWrZLpDG+z/xZaCgQITkA/Dv4V/T6bw7VON6l1Xz/VnrBqrYjZQ12TamKHzITTfOEIYUj48y2KXImA==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/parse-css-color": {
+      "version": "0.2.1",
+      "resolved": "https://registry.npmjs.org/parse-css-color/-/parse-css-color-0.2.1.tgz",
+      "integrity": "sha512-bwS/GGIFV3b6KS4uwpzCFj4w297Yl3uqnSgIPsoQkx7GMLROXfMnWvxfNkL0oh8HVhZA4hvJoEoEIqonfJ3BWg==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "color-name": "^1.1.4",
+        "hex-rgb": "^4.1.0"
+      }
+    },
     "node_modules/parse-gitignore": {
       "version": "2.0.0",
       "resolved": "https://registry.npmjs.org/parse-gitignore/-/parse-gitignore-2.0.0.tgz",
@@ -8882,6 +9288,13 @@
         "node": "^10 || ^12 || >=14"
       }
     },
+    "node_modules/postcss-value-parser": {
+      "version": "4.2.0",
+      "resolved": "https://registry.npmjs.org/postcss-value-parser/-/postcss-value-parser-4.2.0.tgz",
+      "integrity": "sha512-1NNCs6uurfkVbeXG4S8JFT9t19m45ICnif8zWLd5oPSZ50QnwMfK+H3jv408d4jw/7Bttv5axS5IiHoLaVNHeQ==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/postcss-values-parser": {
       "version": "6.0.2",
       "resolved": "https://registry.npmjs.org/postcss-values-parser/-/postcss-values-parser-6.0.2.tgz",
@@ -9423,6 +9836,31 @@
         "node": ">=10"
       }
     },
+    "node_modules/satori": {
+      "version": "0.33.4",
+      "resolved": "https://registry.npmjs.org/satori/-/satori-0.33.4.tgz",
+      "integrity": "sha512-dCnaW6D/tkCJxG5UvpZPWnnvfTdGTVVSAq1tzH6xC5MeOhAeeswRcSq5zT7C6lRCdTGow9PRGus7PCzdp5JJYw==",
+      "dev": true,
+      "license": "MPL-2.0",
+      "dependencies": {
+        "@shuding/opentype.js": "1.4.0-beta.0",
+        "css-background-parser": "^0.1.0",
+        "css-box-shadow": "1.0.0-3",
+        "css-gradient-parser": "^0.0.17",
+        "css-to-react-native": "^3.0.0",
+        "emoji-regex-xs": "^2.0.1",
+        "escape-html": "^1.0.3",
+        "fflate": "0.7.3",
+        "harfbuzzjs": "0.10.0",
+        "linebreak": "^1.1.0",
+        "parse-css-color": "^0.2.1",
+        "postcss-value-parser": "^4.2.0",
+        "yoga-layout": "^3.2.1"
+      },
+      "engines": {
+        "node": ">=16"
+      }
+    },
     "node_modules/sax": {
       "version": "1.6.1",
       "resolved": "https://registry.npmjs.org/sax/-/sax-1.6.1.tgz",
@@ -9855,6 +10293,13 @@
         "node": ">=8"
       }
     },
+    "node_modules/string.prototype.codepointat": {
+      "version": "0.2.1",
+      "resolved": "https://registry.npmjs.org/string.prototype.codepointat/-/string.prototype.codepointat-0.2.1.tgz",
+      "integrity": "sha512-2cBVCj6I4IOvEnjgO/hWqXjqBGsY+zwPmHl12Srk9IXSZ56Jwwmy+66XO5Iut/oQVR7t5ihYdLB0GMa4alEUcg==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/string.prototype.trim": {
       "version": "1.2.11",
       "resolved": "https://registry.npmjs.org/string.prototype.trim/-/string.prototype.trim-1.2.11.tgz",
@@ -10120,6 +10565,13 @@
       "dev": true,
       "license": "MIT"
     },
+    "node_modules/tiny-inflate": {
+      "version": "1.0.3",
+      "resolved": "https://registry.npmjs.org/tiny-inflate/-/tiny-inflate-1.0.3.tgz",
+      "integrity": "sha512-pkY1fj1cKHb2seWDy0B16HeWyczlJA9/WW3u3c4z/NiWDsO3DOU5D7nhTLE9CF0yXv/QZFY7sEJmj24dK+Rrqw==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/tinyclip": {
       "version": "0.1.15",
       "resolved": "https://registry.npmjs.org/tinyclip/-/tinyclip-0.1.15.tgz",
@@ -10405,6 +10857,17 @@
       "integrity": "sha512-jHGmj2LUuqDcX3hqY12Ql+uhUTn8huuxNZGq7GvtF6bSybzH3aFgedYu/KTzQStEgt1Ra2F3HxadNXsNjb3m3g==",
       "license": "MIT"
     },
+    "node_modules/unicode-trie": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/unicode-trie/-/unicode-trie-2.0.0.tgz",
+      "integrity": "sha512-x7bc76x0bm4prf1VLg79uhAzKw8DVboClSN5VxJuQ+LKDOVEW9CdH+VY7SP+vX7xCYQqzzgQpFqz15zeLvAtZQ==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "pako": "^0.2.5",
+        "tiny-inflate": "^1.0.0"
+      }
+    },
     "node_modules/unicorn-magic": {
       "version": "0.1.0",
       "resolved": "https://registry.npmjs.org/unicorn-magic/-/unicorn-magic-0.1.0.tgz",
@@ -11056,6 +11519,13 @@
         "url": "https://github.com/sponsors/sindresorhus"
       }
     },
+    "node_modules/yoga-layout": {
+      "version": "3.2.1",
+      "resolved": "https://registry.npmjs.org/yoga-layout/-/yoga-layout-3.2.1.tgz",
+      "integrity": "sha512-0LPOt3AxKqMdFBZA3HBAt/t/8vIKq7VaQYbuA8WxCgung+p9TVyKRYdpvCb80HcdTN2NkbIKbhNwKUfm3tQywQ==",
+      "dev": true,
+      "license": "MIT"
+    },
     "node_modules/zip-stream": {
       "version": "7.0.5",
       "resolved": "https://registry.npmjs.org/zip-stream/-/zip-stream-7.0.5.tgz",
diff --git a/package.json b/package.json
index 2f6e1cf..1c8c435 100644
--- a/package.json
+++ b/package.json
@@ -8,7 +8,7 @@
   "scripts": {
     "dev": "vite dev --port 3000",
     "build": "vite build",
-    "postbuild": "pagefind --site dist/client",
+    "postbuild": "node scripts/og.ts && pagefind --site dist/client",
     "preview": "vite preview",
     "check": "node scripts/check-build.mjs",
     "test": "node --test \"tests/**/*.test.ts\"",
@@ -27,13 +27,16 @@
   },
   "devDependencies": {
     "@netlify/vite-plugin-tanstack-start": "1.3.19",
+    "@resvg/resvg-js": "^2.6.2",
     "@types/node": "22.12.0",
     "@types/react": "19.2.0",
     "@types/react-dom": "19.2.0",
     "@vitejs/plugin-react": "6.1.1",
     "@vitejs/plugin-rsc": "0.5.34",
+    "jpeg-js": "^0.4.4",
     "pagefind": "1.5.2",
     "prettier": "^3.3.0",
+    "satori": "^0.33.4",
     "typescript": "6.0.3",
     "vite": "8.2.2"
   }
diff --git a/public/artwork.jpg b/public/artwork.jpg
new file mode 100644
index 0000000..21fd1a1
Binary files /dev/null and b/public/artwork.jpg differ
diff --git a/public/robots.txt b/public/robots.txt
new file mode 100644
index 0000000..1996acf
--- /dev/null
+++ b/public/robots.txt
@@ -0,0 +1,4 @@
+User-agent: *
+Allow: /
+
+Sitemap: https://thismonthinreact.com/sitemap.xml
diff --git a/public/styles.css b/public/styles.css
index c8e48a6..697f1d7 100644
--- a/public/styles.css
+++ b/public/styles.css
@@ -1,18 +1,44 @@
 @layer reset, tokens, layout, components, utilities;
 
 /* Cross-document, and outside every layer: @view-transition is ignored inside one. */
-@view-transition { navigation: auto; }
+@view-transition {
+  navigation: auto;
+}
 @media (prefers-reduced-motion: reduce) {
-  @view-transition { navigation: none; }
+  @view-transition {
+    navigation: none;
+  }
 }
 
 @layer reset {
-  *, *::before, *::after { box-sizing: border-box; }
-  * { margin: 0; }
-  img, video, audio { display: block; max-inline-size: 100%; }
-  button, input { font: inherit; color: inherit; }
-  ul, ol { list-style: none; padding: 0; }
-  a { color: inherit; text-underline-offset: 0.15em; }
+  *,
+  *::before,
+  *::after {
+    box-sizing: border-box;
+  }
+  * {
+    margin: 0;
+  }
+  img,
+  video,
+  audio {
+    display: block;
+    max-inline-size: 100%;
+  }
+  button,
+  input {
+    font: inherit;
+    color: inherit;
+  }
+  ul,
+  ol {
+    list-style: none;
+    padding: 0;
+  }
+  a {
+    color: inherit;
+    text-underline-offset: 0.15em;
+  }
 }
 
 @layer tokens {
@@ -27,13 +53,17 @@
     --measure: 70ch;
     --gap: 1.25rem;
     --radius: 0.5rem;
-    --font: ui-sans-serif, system-ui, -apple-system, "Segoe UI", Roboto, sans-serif;
+    --font:
+      ui-sans-serif, system-ui, -apple-system, "Segoe UI", Roboto, sans-serif;
     --mono: ui-monospace, "SF Mono", Menlo, monospace;
   }
 }
 
 @layer layout {
-  html { scroll-padding-block: calc(var(--header-block-size, 4.5rem) + var(--gap)) 6rem; }
+  html {
+    scroll-padding-block: calc(var(--header-block-size, 4.5rem) + var(--gap))
+      6rem;
+  }
 
   body {
     background: var(--bg);
@@ -44,11 +74,28 @@
     padding-inline: var(--gap);
   }
 
-  .site-header, .site-footer, main {
+  .site-header,
+  .site-footer,
+  main {
     max-inline-size: 78rem;
     margin-inline: auto;
   }
 
+  /* Off-screen until focused, so keyboard users can jump the nav. */
+  .skip-link {
+    position: absolute;
+    inset-block-start: 0;
+    inset-inline-start: 0;
+    z-index: 4;
+    padding: 0.5rem 0.75rem;
+    background: var(--bg);
+    border: 1px solid var(--accent);
+    translate: 0 -110%;
+  }
+  .skip-link:focus-visible {
+    translate: 0;
+  }
+
   .site-header {
     display: flex;
     flex-wrap: wrap;
@@ -59,13 +106,31 @@
     border-block-end: 1px solid var(--rule);
   }
 
-  .site-nav { display: flex; flex-wrap: wrap; gap: var(--gap); }
-  .site-name { font-weight: 700; text-decoration: none; letter-spacing: -0.01em; }
-  .site-footer { padding-block: 2rem; color: var(--muted); }
-  main { padding-block: 2rem; display: grid; grid-template-columns: minmax(0, 1fr); gap: 2rem; }
+  .site-nav {
+    display: flex;
+    flex-wrap: wrap;
+    gap: var(--gap);
+  }
+  .site-name {
+    font-weight: 700;
+    text-decoration: none;
+    letter-spacing: -0.01em;
+  }
+  .site-footer {
+    padding-block: 2rem;
+    color: var(--muted);
+  }
+  main {
+    padding-block: 2rem;
+    display: grid;
+    grid-template-columns: minmax(0, 1fr);
+    gap: 2rem;
+  }
 
   /* Container query: outline beside the transcript when the page is wide enough. */
-  .episode-page-main { container: episode / inline-size; }
+  .episode-page-main {
+    container: episode / inline-size;
+  }
 
   /* Title + player ride along at the top; --header-block-size is measured at
      runtime so scroll anchors land below it. */
@@ -82,7 +147,9 @@
     padding-block: 0.6rem;
   }
   @container episode (inline-size > 40rem) {
-    .episode-header { grid-template-columns: minmax(0, 1fr) minmax(16rem, 24rem); }
+    .episode-header {
+      grid-template-columns: minmax(0, 1fr) minmax(16rem, 24rem);
+    }
   }
 
   /* Narrow: cap the outline so the transcript is reachable without a long scroll. */
@@ -93,16 +160,25 @@
     padding-block-end: 0.5rem;
     border-block-end: 1px solid var(--rule);
   }
-  .episode { display: grid; grid-template-columns: minmax(0, 1fr); gap: var(--gap) 2.5rem; align-items: start; }
+  .episode {
+    display: grid;
+    grid-template-columns: minmax(0, 1fr);
+    gap: var(--gap) 2.5rem;
+    align-items: start;
+  }
 
   @container episode (inline-size > 56rem) {
-    .episode { grid-template-columns: 18rem minmax(0, 1fr); }
+    .episode {
+      grid-template-columns: 18rem minmax(0, 1fr);
+    }
     .toc {
       position: sticky;
       /* anchor(--episode-header bottom) computes to `auto` on a sticky box in
          Chrome 152, so the header's measured height carries the offset. */
       inset-block-start: calc(var(--header-block-size, 4.5rem) + var(--gap));
-      max-block-size: calc(100svh - var(--header-block-size, 4.5rem) - 2 * var(--gap));
+      max-block-size: calc(
+        100svh - var(--header-block-size, 4.5rem) - 2 * var(--gap)
+      );
       font-size: 0.9rem;
       border-block-end: 0;
       padding-block-end: 0;
@@ -111,17 +187,41 @@
 }
 
 @layer components {
-  h1, h2, h3 { text-wrap: balance; line-height: 1.2; letter-spacing: -0.015em; margin-block-end: 0.4em; }
-  h1 { font-size: clamp(1.6rem, 1.1rem + 2vw, 2.4rem); }
-  h2 { font-size: 1.25rem; }
-  h3 { font-size: 1rem; }
-  p, li { text-wrap: pretty; }
+  h1,
+  h2,
+  h3 {
+    text-wrap: balance;
+    line-height: 1.2;
+    letter-spacing: -0.015em;
+    margin-block-end: 0.4em;
+  }
+  h1 {
+    font-size: clamp(1.6rem, 1.1rem + 2vw, 2.4rem);
+  }
+  h2 {
+    font-size: 1.25rem;
+  }
+  h3 {
+    font-size: 1rem;
+  }
+  p,
+  li {
+    text-wrap: pretty;
+  }
   /* The reset zeroes margins; restore rhythm without doubling .segment/.link-row. */
-  * + :is(h2, h3) { margin-block-start: 1.5em; }
-  * + p { margin-block-start: 0.75em; }
-  main p { max-inline-size: var(--measure); }
+  * + :is(h2, h3) {
+    margin-block-start: 1.5em;
+  }
+  * + p {
+    margin-block-start: 0.75em;
+  }
+  main p {
+    max-inline-size: var(--measure);
+  }
 
-  .transcript { max-inline-size: var(--measure); }
+  .transcript {
+    max-inline-size: var(--measure);
+  }
 
   .player {
     background: var(--surface);
@@ -129,17 +229,20 @@
     border-radius: var(--radius);
     padding: 0.5rem;
   }
-  .player audio { inline-size: 100%; }
+  .player audio {
+    inline-size: 100%;
+  }
 
   .episode-header h1 {
     font-size: 1.15rem;
     margin-block-end: 0;
-    white-space: nowrap;
-    overflow: hidden;
-    text-overflow: ellipsis;
+    text-wrap: pretty;
   }
 
-  .outline { display: grid; gap: 0.35rem; }
+  .outline {
+    display: grid;
+    gap: 0.35rem;
+  }
   /* Set by the outline script as sections enter the viewport. */
   .outline li:has(> a[aria-current="true"]) {
     border-inline-start: 2px solid var(--accent);
@@ -147,10 +250,24 @@
     margin-inline-start: -0.5rem;
     padding-inline-start: 0.5rem;
   }
-  .outline ol { padding-inline-start: 1rem; margin-block-start: 0.35rem; }
-  .outline a { text-decoration-color: var(--rule); }
-  .outbound { color: var(--muted); text-decoration: none; }
-  .toc h2 { font-size: 0.8rem; text-transform: uppercase; letter-spacing: 0.08em; color: var(--muted); margin-block-end: 0.6rem; }
+  .outline ol {
+    padding-inline-start: 1rem;
+    margin-block-start: 0.35rem;
+  }
+  .outline a {
+    text-decoration-color: var(--rule);
+  }
+  .outbound {
+    color: var(--muted);
+    text-decoration: none;
+  }
+  .toc h2 {
+    font-size: 0.8rem;
+    text-transform: uppercase;
+    letter-spacing: 0.08em;
+    color: var(--muted);
+    margin-block-end: 0.6rem;
+  }
 
   .ts {
     font-family: var(--mono);
@@ -161,54 +278,142 @@
     border: 0;
     padding: 0;
   }
-  button.ts { cursor: pointer; }
-  button.ts:hover, button.ts:focus-visible { text-decoration: underline; }
+  button.ts {
+    cursor: pointer;
+  }
+  button.ts:hover,
+  button.ts:focus-visible {
+    text-decoration: underline;
+  }
 
-  .transcript section + section { margin-block-start: 2.5rem; }
-  .transcript h2 { padding-block-end: 0.4rem; border-block-end: 1px solid var(--rule); }
-  .segment { margin-block-start: 0.85em; }
-  .speaker { color: var(--accent); font-weight: 600; }
+  .transcript section + section {
+    margin-block-start: 2.5rem;
+  }
+  .transcript h2 {
+    padding-block-end: 0.4rem;
+    border-block-end: 1px solid var(--rule);
+  }
+  .segment {
+    margin-block-start: 0.85em;
+  }
+  .speaker {
+    color: var(--accent);
+    font-weight: 600;
+  }
 
-  .latest h1 a { text-decoration: none; }
-  .archive ul, .link-group ul, .people ul, .subscribe ul, .replies { display: grid; gap: 0.9rem; }
-  .archive li { padding-block-end: 0.9rem; border-block-end: 1px solid var(--rule); }
-  .archive time, .from, .role, .count, .reply-date { color: var(--muted); font-size: 0.85em; }
-  .archive time { display: block; font-family: var(--mono); }
-  .count { font-family: var(--mono); font-weight: 400; }
+  .latest h1 a {
+    text-decoration: none;
+  }
+  .archive ul,
+  .link-group ul,
+  .people ul,
+  .subscribe ul,
+  .replies {
+    display: grid;
+    gap: 0.9rem;
+  }
+  .archive li {
+    padding-block-end: 0.9rem;
+    border-block-end: 1px solid var(--rule);
+  }
+  .archive time,
+  .from,
+  .role,
+  .count,
+  .reply-date {
+    color: var(--muted);
+    font-size: 0.85em;
+  }
+  .archive time {
+    display: block;
+    font-family: var(--mono);
+  }
+  .count {
+    font-family: var(--mono);
+    font-weight: 400;
+  }
 
-  #link-filter, .newsletter input {
+  #link-filter,
+  .newsletter input {
     inline-size: min(100%, 24rem);
     padding: 0.5rem 0.7rem;
     border: 1px solid var(--rule);
     border-radius: var(--radius);
     background: var(--surface);
   }
-  #link-count { color: var(--muted); font-size: 0.85em; }
-  .link-group { margin-block-start: 1.5rem; }
-  .link-row { display: flex; flex-wrap: wrap; gap: 0.25rem 0.75rem; align-items: baseline; }
-  .link-row > a:first-child { flex: 1 1 22rem; }
-  .from { text-decoration: none; }
-  .from .ts { margin-inline-start: 0.4rem; }
+  #link-count {
+    color: var(--muted);
+    font-size: 0.85em;
+  }
+  .link-group {
+    margin-block-start: 1.5rem;
+  }
+  .link-row {
+    display: flex;
+    flex-wrap: wrap;
+    gap: 0.25rem 0.75rem;
+    align-items: baseline;
+  }
+  .link-row > a:first-child {
+    flex: 1 1 22rem;
+  }
+  .from {
+    text-decoration: none;
+  }
+  .from .ts {
+    margin-inline-start: 0.4rem;
+  }
 
-  .people li { display: flex; gap: 0.7rem; align-items: center; }
-  .people img { border-radius: 50%; }
-  .about-page section, #comments { margin-block-start: 2rem; }
+  .people li {
+    display: flex;
+    gap: 0.7rem;
+    align-items: center;
+  }
+  .people img {
+    border-radius: 50%;
+  }
+  .about-page section,
+  #comments {
+    margin-block-start: 2rem;
+  }
 
-  #comments .replies { margin-block-start: 1rem; }
-  .replies li { border-inline-start: 2px solid var(--rule); padding-inline-start: 0.8rem; }
-  .reply-author { font-weight: 600; font-size: 0.9em; }
-  .reply-text { max-inline-size: var(--measure); }
+  #comments .replies {
+    margin-block-start: 1rem;
+  }
+  .replies li {
+    border-inline-start: 2px solid var(--rule);
+    padding-inline-start: 0.8rem;
+  }
+  .reply-author {
+    font-weight: 600;
+    font-size: 0.9em;
+  }
+  .reply-text {
+    max-inline-size: var(--measure);
+  }
 
   .newsletter form {
-    display: flex; flex-wrap: wrap; gap: 0.5rem;
-    align-items: center; margin-block-start: 0.7rem;
+    display: flex;
+    flex-wrap: wrap;
+    gap: 0.5rem;
+    align-items: center;
+    margin-block-start: 0.7rem;
+  }
+  .newsletter input {
+    flex: 1 1 16rem;
   }
-  .newsletter input { flex: 1 1 16rem; }
   .newsletter button {
-    padding: 0.5rem 1rem; border: 0; border-radius: var(--radius);
-    background: var(--accent); color: var(--bg); cursor: pointer;
+    padding: 0.5rem 1rem;
+    border: 0;
+    border-radius: var(--radius);
+    background: var(--accent);
+    color: var(--bg);
+    cursor: pointer;
+  }
+  .newsletter label {
+    position: absolute;
+    clip-path: inset(50%);
   }
-  .newsletter label { position: absolute; clip-path: inset(50%); }
 
   /* Pagefind ships its own CSS; only its documented custom properties are set. */
   #pagefind-ui {
@@ -223,14 +428,23 @@
 }
 
 @layer utilities {
-  [hidden] { display: none !important; }
-  :focus-visible { outline: 2px solid var(--accent); outline-offset: 2px; }
+  [hidden] {
+    display: none !important;
+  }
+  :focus-visible {
+    outline: 2px solid var(--accent);
+    outline-offset: 2px;
+  }
 
   @media (prefers-reduced-motion: reduce) {
-    *, *::before, *::after {
+    *,
+    *::before,
+    *::after {
       animation-duration: 0.01ms !important;
       transition-duration: 0.01ms !important;
     }
-    .toc { scroll-behavior: auto; }
+    .toc {
+      scroll-behavior: auto;
+    }
   }
 }
diff --git a/scripts/check-build.mjs b/scripts/check-build.mjs
index 9c990b4..962d338 100644
--- a/scripts/check-build.mjs
+++ b/scripts/check-build.mjs
@@ -32,17 +32,26 @@ function contains(path, needle) {
   );
 }
 
-console.log("Task 1: shell");
+console.log("Shell");
 hasFile("index.html");
 hasFile("styles.css");
 contains("index.html", 'class="site-nav"');
 contains("index.html", 'href="/styles.css"');
+// The skip link must precede the header on both rendering paths — the router
+// shell and the static document handlers build their own markup.
+for (const path of ["index.html", "links/index.html"]) {
+  const html = file(path) || "";
+  check(
+    `${path} has a skip link before the header`,
+    html.includes('class="skip-link" href="#main"') &&
+      html.includes('id="main"') &&
+      html.indexOf("skip-link") < html.indexOf("site-header"),
+  );
+}
 
-// --- Task 4: home route (appended) ---
 contains("index.html", 'class="archive"');
 contains("index.html", 'class="outline"');
 for (const slug of slugs()) contains("index.html", `/episodes/${slug}`);
-// --- end Task 4 ---
 
 function slugs() {
   return readdirSync("content/episodes")
@@ -51,7 +60,7 @@ function slugs() {
     .sort();
 }
 
-console.log("\nTask 3: episode documents");
+console.log("\nEpisode documents");
 const episodeSlugs = slugs();
 check("content/episodes has at least one episode", episodeSlugs.length > 0);
 for (const slug of episodeSlugs) hasFile(`episodes/${slug}/index.html`);
@@ -64,27 +73,7 @@ if (newest) {
   contains(path, "<audio");
   contains(path, 'class="player"');
   contains(path, 'property="og:title"');
-  const html = file(path) || "";
-  check(
-    `${path} references no bundled JS`,
-    !/\/assets\/[^"']*\.js/.test(html),
-    "static documents must ship zero external scripts",
-  );
-  // The transcript must appear exactly once — a second copy means a loader or
-  // RSC payload has been reintroduced.
   const md = readFileSync(join("content/episodes", `${newest}.md`), "utf8");
-  const phrase = (md.split("\n# Transcript\n")[1] || "")
-    .split("\n")
-    .map((line) => /[A-Za-z][A-Za-z ]{29,59}[A-Za-z]/.exec(line)?.[0])
-    .find(Boolean);
-  if (phrase) {
-    const copies = (file(path) || "").split(phrase).length - 1;
-    check(
-      `${path} contains the transcript exactly once`,
-      copies === 1,
-      `found ${copies}`,
-    );
-  }
 
   // AT Protocol bits are conditional on front matter and env, so only assert
   // them when the inputs are actually present.
@@ -103,8 +92,7 @@ if (newest) {
   }
 }
 
-// --- Task 5: links document (appended) ---
-console.log("\nTask 5: links document");
+console.log("\nLinks document");
 hasFile("links/index.html");
 contains("links/index.html", 'class="link-row"');
 contains("links/index.html", "data-text=");
@@ -113,9 +101,8 @@ check(
   "links/index.html references no bundled JS",
   !/\/assets\/[^"']*\.js/.test(file("links/index.html") || ""),
 );
-// --- end Task 5 ---
 
-console.log("\nTask 6: feed and chapters");
+console.log("\nFeed and chapters");
 hasFile("feed.xml");
 contains("feed.xml", '<rss version="2.0"');
 contains("feed.xml", "<item>");
@@ -152,7 +139,7 @@ for (const slug of slugs()) {
   );
 }
 
-console.log("\nTask 7: search");
+console.log("\nSearch");
 hasFile("search/index.html");
 contains("search/index.html", 'id="pagefind-ui"');
 check(
@@ -180,7 +167,7 @@ if (existsSync(join(dist, "pagefind", "pagefind-ui.js"))) {
   }
 }
 
-console.log("\nTask 8: about");
+console.log("\nAbout");
 hasFile("about/index.html");
 contains("about/index.html", 'class="people"');
 contains(
@@ -210,7 +197,7 @@ if (process.env.VITE_BLUESKY_PROFILE_URL)
   );
 }
 
-console.log("\nTask 9: stylesheet");
+console.log("\nStylesheet");
 const cssSource = readFileSync("public/styles.css", "utf8");
 for (const needle of [
   "@layer reset, tokens, layout, components, utilities;",
@@ -224,11 +211,9 @@ for (const needle of [
 ]) {
   check(`styles.css uses ${needle}`, cssSource.includes(needle));
 }
-const lineCount = cssSource.split("\n").length;
-check("styles.css is under 250 lines", lineCount < 250, `${lineCount} lines`);
 check("styles.css is served verbatim", file("styles.css") === cssSource);
 
-console.log("\nTask 10: prerender coverage");
+console.log("\nPrerender coverage");
 const allSlugs = slugs();
 const expected = [
   "index.html",
@@ -242,13 +227,13 @@ const expected = [
     `episodes/${s}/chapters.json`,
   ]),
 ];
-for (const path of expected) hasFile(path);
 check(
   `prerendered all ${expected.length} expected outputs`,
   expected.every((p) => file(p) !== null),
+  `missing: ${expected.filter((p) => file(p) === null).join(", ")}`,
 );
 
-console.log("\nTask 10: standard.site discovery");
+console.log("\nstandard.site discovery");
 const wellKnown = ".well-known/site.standard.publication";
 if (process.env.VITE_ATPROTO_PUBLICATION_URI) {
   hasFile(wellKnown);
@@ -261,7 +246,7 @@ if (process.env.VITE_ATPROTO_PUBLICATION_URI) {
   console.log(`  skip  ${wellKnown} (VITE_ATPROTO_PUBLICATION_URI unset)`);
 }
 
-console.log("\nTask 10: no duplicated transcript");
+console.log("\nTranscript integrity");
 for (const slug of allSlugs) {
   const html = file(`episodes/${slug}/index.html`);
   if (html === null) continue;
@@ -286,8 +271,7 @@ for (const slug of allSlugs) {
   );
 }
 
-// --- Adjustments: sticky episode header + runtime outline highlighting ---
-console.log("\nAdjustments: episode header and outline state");
+console.log("\nEpisode header and outline state");
 for (const slug of allSlugs) {
   const path = `episodes/${slug}/index.html`;
   const html = file(path);
@@ -300,7 +284,84 @@ for (const slug of allSlugs) {
     "the outline script must set it at runtime only",
   );
 }
-// --- end Adjustments ---
+
+console.log("\nOpen Graph cards");
+for (const slug of allSlugs) {
+  check(
+    `og/${slug}.jpg exists`,
+    existsSync(join(dist, "og", `${slug}.jpg`)),
+    "run: node scripts/og.ts",
+  );
+}
+check("og/default.jpg exists", existsSync(join(dist, "og", "default.jpg")));
+
+console.log("\nLink previews");
+const origin = (
+  process.env.VITE_SITE_URL || "https://thismonthinreact.com"
+).replace(/\/$/, "");
+for (const [path, slug] of [
+  ["index.html", "default"],
+  ["about/index.html", "default"],
+  [`episodes/${allSlugs.at(-1)}/index.html`, allSlugs.at(-1)],
+]) {
+  contains(path, `content="${origin}/og/${slug}.jpg"`);
+  contains(path, 'content="summary_large_image"');
+  contains(path, `<link rel="canonical"`);
+  contains(path, 'property="og:image:width" content="1200"');
+  contains(path, 'name="theme-color"');
+}
+{
+  const path = `episodes/${allSlugs.at(-1)}/index.html`;
+  contains(path, 'property="og:type" content="article"');
+  contains(path, 'property="article:published_time"');
+  const title = /<title>([^<]*)<\/title>/.exec(file(path) || "")?.[1] || "";
+  check(
+    `${path} title names the site once`,
+    (title.match(/This Month in React/g) || []).length === 1,
+    title,
+  );
+}
+
+console.log("\nRobots, sitemap, JSON-LD");
+hasFile("robots.txt");
+contains("robots.txt", "Sitemap:");
+contains("robots.txt", "/sitemap.xml");
+hasFile("sitemap.xml");
+contains("sitemap.xml", "<urlset");
+for (const slug of allSlugs) contains("sitemap.xml", `/episodes/${slug}<`);
+check(
+  "sitemap.xml omits /search",
+  !(file("sitemap.xml") || "").includes("/search"),
+);
+
+/** The first ld+json block in a document, parsed. */
+function ldJson(path) {
+  const m = /<script type="application\/ld\+json">([\s\S]*?)<\/script>/.exec(
+    file(path) || "",
+  );
+  if (!m) return null;
+  try {
+    return JSON.parse(m[1]);
+  } catch {
+    return null;
+  }
+}
+check(
+  "index.html has PodcastSeries JSON-LD",
+  ldJson("index.html")?.["@type"] === "PodcastSeries",
+);
+{
+  const path = `episodes/${allSlugs.at(-1)}/index.html`;
+  const ld = ldJson(path);
+  check(
+    `${path} has PodcastEpisode JSON-LD`,
+    ld?.["@type"] === "PodcastEpisode",
+  );
+  check(
+    `${path} JSON-LD names its series`,
+    ld?.partOfSeries?.["@type"] === "PodcastSeries",
+  );
+}
 
 console.log("\nSizes");
 const assetsDir = join(dist, "assets");
diff --git a/scripts/ingest.ts b/scripts/ingest.ts
index f45c7c0..1987ae3 100644
--- a/scripts/ingest.ts
+++ b/scripts/ingest.ts
@@ -3,6 +3,7 @@ import { join, resolve } from "node:path";
 import type { Person } from "../src/content/parse.ts";
 import { splitFile } from "../src/content/parse.ts";
 import { serializeEpisodeFile } from "../src/content/serialize.ts";
+import { toSeconds } from "../src/content/time.ts";
 
 export const FEED_URL = "https://feeds.transistor.fm/this-month-in-react";
 
@@ -18,14 +19,26 @@ export interface FeedItem {
 }
 
 const MONTHS = [
-  "january", "february", "march", "april", "may", "june",
-  "july", "august", "september", "october", "november", "december",
+  "january",
+  "february",
+  "march",
+  "april",
+  "may",
+  "june",
+  "july",
+  "august",
+  "september",
+  "october",
+  "november",
+  "december",
 ];
 
 function decode(s: string): string {
   return s
     .replace(/&#(\d+);/g, (_, n) => String.fromCodePoint(Number(n)))
-    .replace(/&#x([0-9a-f]+);/gi, (_, n) => String.fromCodePoint(parseInt(n, 16)))
+    .replace(/&#x([0-9a-f]+);/gi, (_, n) =>
+      String.fromCodePoint(parseInt(n, 16)),
+    )
     .replace(/&lt;/g, "<")
     .replace(/&gt;/g, ">")
     .replace(/&quot;/g, '"')
@@ -89,8 +102,15 @@ export function parseFeed(xml: string): FeedItem[] {
     const transistorId = audioUrl
       ? /media\.transistor\.fm\/([^/?#]+)/.exec(audioUrl)?.[1]
       : undefined;
-    const duration = Number(tag(item, "itunes:duration") ?? NaN);
-    if (!transistorId || !audioUrl || Number.isNaN(duration)) continue;
+    // <itunes:duration> is legally either raw seconds or HH:MM:SS / MM:SS;
+    // toSeconds handles both.
+    const duration = toSeconds(tag(item, "itunes:duration"));
+    if (!transistorId || !audioUrl || duration === undefined) {
+      console.warn(
+        `skipping feed item ${epSlug}: missing audio url or duration`,
+      );
+      continue;
+    }
 
     const people: Person[] = [];
     for (const m of item.matchAll(
@@ -108,9 +128,10 @@ export function parseFeed(xml: string): FeedItem[] {
     const episode = Number(tag(item, "podcast:episode") ?? NaN);
     // The "Reply on Bluesky" anchor inside the <description> CDATA. Matched on
     // the URL shape, not the link text, so a renamed link still resolves.
-    const bskyPostUrl = /href="(https:\/\/bsky\.app\/profile\/[^/"]+\/post\/[^"]+)"/.exec(
-      item,
-    )?.[1];
+    const bskyPostUrl =
+      /href="(https:\/\/bsky\.app\/profile\/[^/"]+\/post\/[^"]+)"/.exec(
+        item,
+      )?.[1];
     items.push({
       slug: epSlug,
       transistorId,
@@ -137,8 +158,11 @@ export function applyFeedItem(fileText: string, item: FeedItem): string {
   frontMatter.duration = item.duration;
   if (item.season !== undefined) frontMatter.season = item.season;
   if (item.episode !== undefined) frontMatter.episode = item.episode;
-  frontMatter.people = item.people;
-  if (item.bskyPostUrl !== undefined) frontMatter.bskyPostUrl = item.bskyPostUrl;
+  // An empty list means the feed item carried no <podcast:person> tags, not
+  // that the episode has no people: never clobber hand-curated front matter.
+  if (item.people.length > 0) frontMatter.people = item.people;
+  if (item.bskyPostUrl !== undefined)
+    frontMatter.bskyPostUrl = item.bskyPostUrl;
   return serializeEpisodeFile(frontMatter, body);
 }
 
diff --git a/scripts/migrate.ts b/scripts/migrate.ts
deleted file mode 100644
index 1bc6621..0000000
--- a/scripts/migrate.ts
+++ /dev/null
@@ -1,338 +0,0 @@
-import { readdirSync, readFileSync, writeFileSync, mkdirSync, existsSync } from "node:fs";
-import { join, resolve } from "node:path";
-import { flattenLinks, normalizeTime, slug } from "../src/content/slug.ts";
-import { splitFile } from "../src/content/parse.ts";
-import { insertHeadings } from "../src/content/headings.ts";
-import { serializeEpisodeFile } from "../src/content/serialize.ts";
-
-const MONTHS = [
-  "january", "february", "march", "april", "may", "june",
-  "july", "august", "september", "october", "november", "december",
-];
-
-/** "tmir-2024-03.md" | "tmir-march-2023.md" | "tmir-dec-2023.md" -> "yyyy-mm" */
-export function slugFromFilename(name: string): string | null {
-  const iso = /^tmir-(\d{4})-(\d{2})\.md$/.exec(name);
-  if (iso) return `${iso[1]}-${iso[2]}`;
-
-  const named = /^tmir-([a-z]+)-(\d{4})\.md$/.exec(name);
-  if (!named) return null;
-  const idx = MONTHS.findIndex((m) => m.startsWith(named[1]));
-  if (idx === -1) return null;
-  return `${named[2]}-${String(idx + 1).padStart(2, "0")}`;
-}
-
-const SPEAKER_PARA = /^(?:\[\d{1,3}(?::\d{2}){1,2}\]\s*)?\*\*[^*]+:\*\*/;
-// Unbolded speaker prefix, e.g. `Carl: ` or a raw Descript id `2-vcarl: `
-// (tmir-2025-09.md, tmir-2025-11.md never wrap the label in `**...**`).
-// Matching on shape alone is too loose (`Note: something` looks identical),
-// so a match here only counts as a speaker if isKnownSpeakerToken() confirms it.
-const UNBOLDED_SPEAKER = /^(?:\[\d{1,3}(?::\d{2}){1,2}\]\s*)?([A-Za-z0-9-]{1,20}):\s+/;
-const LEADING_TIME = /^\[(\d{1,3}(?::\d{2}){1,2})\]\s*/;
-const TRAILING_TIME = /\s*\[(\d{1,3}(?::\d{2}){1,2})\]\s*$/;
-const HEADING_TIME = /^(#{1,2})\s*\[(\d{1,3}(?::\d{2}){1,2})\]\s*(.*)$/;
-const OUTLINE_LINE = /^(\s*)- (.+)$/;
-// `- [[mm:ss](#anchor)] Title` (linked) or a bare `- [mm:ss] Title` (2023-10,
-// 2023-11 have no anchor link at all). Both may sit behind a short prefix
-// like `⚡️ ` (2024-04's lightning-round bullets); PREFIX captures that.
-const PREFIX = /^([^[]{1,4})(?=\[)/;
-const OUTLINE_TIME_LINKED = /^\[\[(\d{1,3}(?::\d{2}){1,2})\]\(#[^)]*\)\]\s*/;
-const OUTLINE_TIME_BARE = /^\[(\d{1,3}(?::\d{2}){1,2})\]\s*/;
-
-// Short first-name / raw-Descript-id -> full name, for the unbolded-speaker
-// files where the label isn't the canonical "Carl Vitullo" form the bolded
-// files use. Descript ids look like `1-vcarl`; strip the leading `N-`.
-const NAME_MAP: Record<string, string> = {
-  carl: "Carl Vitullo",
-  vcarl: "Carl Vitullo",
-  mark: "Mark Erikson",
-  acemarke: "Mark Erikson",
-  mo: "Mo Javad",
-  mojavad: "Mo Javad",
-};
-
-/**
- * Canonicalize a speaker label, bolded or not. A trailing parenthetical is an
- * editing note, not part of the name (`Carl Vitullo (editing)`), so it's
- * dropped before the lookup. Bare `Mo` is Mo Javad except in the files where
- * Mo Khazali is the guest — there the short form is his, so leave it alone.
- * Anything not in the map (guests) keeps the name it already has.
- */
-export function normalizeSpeakerName(raw: string, boldedNames?: Set<string>): string {
-  const name = raw.replace(/\s*\([^)]*\)\s*$/, "").trim();
-  const key = name.replace(/^\d+-/, "").toLowerCase();
-  if (key === "mo" && boldedNames?.has("mo khazali")) return name;
-  return NAME_MAP[key] ?? name;
-}
-
-/** Every name (and its first word) that appears bolded as a speaker anywhere in the file. */
-function collectBoldedSpeakers(body: string): Set<string> {
-  const names = new Set<string>();
-  for (const m of body.matchAll(/\*\*([^*]+):\*\*/g)) {
-    const name = m[1].trim().toLowerCase();
-    names.add(name);
-    names.add(name.split(/\s+/)[0]);
-  }
-  return names;
-}
-
-/** Is `raw` (an UNBOLDED_SPEAKER match) an actual known speaker, not just `Word:` prose? */
-function isKnownSpeakerToken(raw: string, boldedNames: Set<string>): boolean {
-  const key = raw.replace(/^\d+-/, "").toLowerCase();
-  return key in NAME_MAP || boldedNames.has(key);
-}
-
-/** Remove <style>…</style> and <iframe …></iframe> blocks. */
-function stripEmbeds(body: string): string {
-  return body
-    .replace(/<style>[\s\S]*?<\/style>\s*/g, "")
-    .replace(/<iframe[\s\S]*?<\/iframe>\s*/g, "");
-}
-
-export function convert(raw: string): { text: string; warnings: string[] } {
-  const warnings: string[] = [];
-  const { frontMatter, body } = splitFile(raw);
-  const clean = stripEmbeds(body);
-  const blocks = clean.split(/\n{2,}/).map((b) => b.trim()).filter(Boolean);
-  const boldedNames = collectBoldedSpeakers(clean);
-
-  function isSpeakerParagraph(block: string): boolean {
-    if (SPEAKER_PARA.test(block)) return true;
-    const u = UNBOLDED_SPEAKER.exec(block);
-    return !!u && isKnownSpeakerToken(u[1], boldedNames);
-  }
-
-  // The transcript starts at the first speaker paragraph, plus any `##`
-  // section headings sitting directly above it. Walk back over `##` only:
-  // a lone `#` there is the old transcript-start marker (`# Interview`), and
-  // a `#` elsewhere in the region is outline structure (2024-04's
-  // `# Main Content`).
-  let start = blocks.findIndex((b) => isSpeakerParagraph(b));
-  if (start === -1) {
-    warnings.push("no speaker paragraphs found; body left in the outline region");
-  }
-  while (start > 0 && blocks[start - 1].startsWith("## ")) start -= 1;
-
-  const head = start === -1 ? blocks : blocks.slice(0, start);
-  const tail = start === -1 ? [] : blocks.slice(start);
-
-  // --- outline ---
-  const outlineLines: string[] = [];
-  // Top-level (non-nested) outline items with a time, in source order — used
-  // below to synthesize section headings for a transcript with none at all.
-  const topLevelItems: { time: string; title: string }[] = [];
-  for (const [i, block] of head.entries()) {
-    if (!OUTLINE_LINE.test(block.split("\n")[0])) {
-      const isMarker = i === head.length - 1 && /^#\s/.test(block);
-      if (isMarker) continue; // old transcript-start marker; drop silently
-      if (block.startsWith("#")) warnings.push(`dropped heading before outline: ${block.split("\n")[0]}`);
-      else warnings.push(`dropped prose before outline: ${block.slice(0, 60)}`);
-      continue;
-    }
-    for (const line of block.split("\n")) {
-      const m = OUTLINE_LINE.exec(line);
-      if (!m) continue;
-      let rest = m[2].trim();
-      // Drop the pre-iframe subscribe link list.
-      if (/^\[(Spotify|Apple Podcasts|RSS|or anywhere you prefer)\]\(/.test(rest)) continue;
-
-      // A short prefix (e.g. "⚡️ ") may sit before the timestamp; keep it
-      // in the title and just skip past it to find the time.
-      const p = PREFIX.exec(rest);
-      const prefix = p ? p[1] : "";
-      const afterPrefix = rest.slice(prefix.length);
-
-      let time: string | undefined;
-      const linked = OUTLINE_TIME_LINKED.exec(afterPrefix);
-      const bare = linked ? null : OUTLINE_TIME_BARE.exec(afterPrefix);
-      const match = linked ?? bare;
-      if (match) {
-        time = normalizeTime(match[1]);
-        rest = prefix + afterPrefix.slice(match[0].length);
-      } else {
-        warnings.push(`outline item without timestamp: ${rest.slice(0, 60)}`);
-      }
-      const anchor = slug(flattenLinks(rest).trim());
-      const timeMarker = time ? `[[${time}](#${anchor})] ` : "";
-      outlineLines.push(`${m[1]}- ${timeMarker}${rest}`);
-      if (m[1].length === 0 && time) topLevelItems.push({ time, title: rest });
-    }
-  }
-
-  // --- transcript ---
-  type Block = { text: string; time?: string; isHeading: boolean };
-  const out: Block[] = [];
-  const sectionTimes = new Map<string, string>();
-  let pendingHeadingAnchor: string | null = null;
-
-  for (const block of tail) {
-    const h = HEADING_TIME.exec(block);
-    if (h) {
-      // `## [00:52] job market` -> heading loses the timestamp
-      const title = h[3].trim();
-      const anchor = slug(title);
-      sectionTimes.set(anchor, normalizeTime(h[2]));
-      pendingHeadingAnchor = anchor;
-      out.push({ text: `## ${title}`, isHeading: true });
-      continue;
-    }
-    if (block.startsWith("#")) {
-      const level = /^#+/.exec(block)![0].length;
-      const title = block.replace(/^#+\s*/, "").trim();
-      // 2023 used h1 for sections; promote to h2.
-      pendingHeadingAnchor = slug(title);
-      out.push({ text: `## ${title}`, isHeading: true });
-      if (level > 2) warnings.push(`heading deeper than h2 flattened: ${title}`);
-      continue;
-    }
-
-    // Paragraph: move a leading timestamp to the end.
-    let text = block;
-    const lead = LEADING_TIME.exec(text);
-    let time: string | undefined;
-    if (lead) {
-      time = normalizeTime(lead[1]);
-      text = text.slice(lead[0].length);
-    } else {
-      const trail = TRAILING_TIME.exec(text);
-      if (trail) {
-        time = normalizeTime(trail[1]);
-        text = text.slice(0, trail.index);
-      }
-    }
-    // Speaker label -> bold + canonical name. Bolded labels are normalized
-    // too: the sources mix `**Carl:**`, `**Carl Vitullo:**` and
-    // `**Carl (editing):**` for the same person.
-    const b = /^\*\*([^*]+):\*\*\s*/.exec(text);
-    if (b) {
-      text = `**${normalizeSpeakerName(b[1], boldedNames)}:** ${text.slice(b[0].length)}`;
-    } else {
-      const u = UNBOLDED_SPEAKER.exec(text);
-      if (u && isKnownSpeakerToken(u[1], boldedNames)) {
-        text = `**${normalizeSpeakerName(u[1], boldedNames)}:** ${text.slice(u[0].length)}`;
-      }
-    }
-
-    if (!time) warnings.push(`paragraph without timestamp: ${text.slice(0, 60)}`);
-    if (!/^\*\*[^*]+:\*\*/.test(text) && out.length === 0) {
-      warnings.push(`paragraph without speaker: ${text.slice(0, 60)}`);
-    }
-    if (pendingHeadingAnchor && time && !sectionTimes.has(pendingHeadingAnchor)) {
-      sectionTimes.set(pendingHeadingAnchor, time);
-      pendingHeadingAnchor = null;
-    }
-    out.push({ text: time ? `${text.trim()} [${time}]` : text.trim(), time, isHeading: false });
-  }
-
-  // A transcript with no section headings at all (2023-10: an AI-written
-  // outline with real timestamps, but a raw Descript export with no `##`
-  // markers) gets headings synthesized from its own top-level outline items.
-  if (!out.some((b) => b.isHeading) && topLevelItems.length > 0) {
-    const merged = insertHeadings(out.map((b) => b.text), topLevelItems);
-    out.length = 0;
-    for (const text of merged) out.push({ text, isHeading: text.startsWith("## ") });
-  }
-
-  // No outline in the source: synthesize one from the section headings.
-  if (outlineLines.length === 0) {
-    for (const block of out) {
-      if (!block.isHeading) continue;
-      const title = block.text.slice(3).trim();
-      const anchor = slug(title);
-      const time = sectionTimes.get(anchor);
-      outlineLines.push(time ? `- [[${time}](#${anchor})] ${title}` : `- ${title}`);
-    }
-  }
-
-  const newBody = `\n${outlineLines.join("\n")}\n\n# Transcript\n\n${out.map((b) => b.text).join("\n\n")}\n`;
-  return { text: serializeEpisodeFile(frontMatter, newBody), warnings };
-}
-
-// Front matter the legacy sources never carry: written by scripts/ingest.ts
-// (everything but atUri) and scripts/publish-atproto.ts (atUri). Regenerating
-// an episode must not throw these away.
-const PRESERVED_FIELDS = [
-  "transistorId",
-  "audioUrl",
-  "duration",
-  "season",
-  "episode",
-  "people",
-  "bskyPostUrl",
-  "atUri",
-] as const;
-
-/**
- * Carry the ingest-owned (and atUri) front matter of an already-migrated file
- * into a freshly regenerated one. Hand-written fields come from the
- * regenerated text; the preserved fields come from the existing file.
- */
-export function mergeExistingFrontMatter(newText: string, existingText: string): string {
-  const fresh = splitFile(newText);
-  const existing = splitFile(existingText);
-  for (const key of PRESERVED_FIELDS) {
-    if (existing.frontMatter[key] !== undefined) {
-      fresh.frontMatter[key] = existing.frontMatter[key];
-    }
-  }
-  return serializeEpisodeFile(fresh.frontMatter, fresh.body);
-}
-
-const SOURCE_DIR = resolve(
-  process.env.TMIR_SOURCE_DIR ??
-    "/Users/vcarl/workspace/reactiflux.com/src/transcripts",
-);
-const OUT_DIR = resolve("content/episodes");
-
-function main() {
-  const force = process.argv.includes("--force");
-  mkdirSync(OUT_DIR, { recursive: true });
-
-  const sources = readdirSync(SOURCE_DIR)
-    .sort()
-    .map((name) => ({ name, epSlug: slugFromFilename(name) }))
-    .filter((s): s is { name: string; epSlug: string } => s.epSlug !== null);
-
-  const existing = sources.filter((s) => existsSync(join(OUT_DIR, `${s.epSlug}.md`)));
-  if (existing.length > 0 && !force) {
-    console.error(
-      `${existing.length} target file(s) already exist and would be overwritten:`,
-    );
-    for (const s of existing) console.error(`    content/episodes/${s.epSlug}.md`);
-    console.error("\nRe-run with --force to regenerate them (ingest-owned front matter is preserved).");
-    process.exit(1);
-  }
-
-  let files = 0;
-  let warned = 0;
-  const skipped: string[] = [];
-  for (const { name, epSlug } of sources) {
-    const outPath = join(OUT_DIR, `${epSlug}.md`);
-    const { text: fresh, warnings } = convert(readFileSync(join(SOURCE_DIR, name), "utf8"));
-    const text = existsSync(outPath)
-      ? mergeExistingFrontMatter(fresh, readFileSync(outPath, "utf8"))
-      : fresh;
-    // No speaker paragraph found means the whole transcript uses a speaker
-    // format this script doesn't recognize (e.g. bare `Carl:` / `2-vcarl:`
-    // without bold markup, as in tmir-2025-09.md and tmir-2025-11.md) — the
-    // entire body would be misfiled as outline and the output would parse
-    // but carry zero transcript sections. Leave it out rather than emit a
-    // silently-empty file.
-    if (warnings.some((w) => w.startsWith("no speaker paragraphs found"))) {
-      console.log(`${name} -> SKIPPED (no recognized speaker paragraphs)`);
-      skipped.push(name);
-      continue;
-    }
-    writeFileSync(outPath, text);
-    files += 1;
-    console.log(`${name} -> content/episodes/${epSlug}.md  (${warnings.length} warnings)`);
-    for (const w of warnings) {
-      console.log(`    ${w}`);
-      warned += 1;
-    }
-  }
-  console.log(`\n${files} files migrated, ${warned} warnings, ${skipped.length} skipped.`);
-  if (skipped.length) console.log(`skipped: ${skipped.join(", ")}`);
-}
-
-if (import.meta.filename === process.argv[1]) main();
diff --git a/scripts/og.ts b/scripts/og.ts
new file mode 100644
index 0000000..dc31b1c
--- /dev/null
+++ b/scripts/og.ts
@@ -0,0 +1,180 @@
+/**
+ * Build-time Open Graph cards: one 1200x630 JPEG per episode at
+ * dist/client/og/<slug>.jpg, plus dist/client/og/default.jpg for every other
+ * page. Runs offline — fonts and artwork are committed in this repo.
+ */
+import { mkdir, readFile, writeFile } from "node:fs/promises";
+import { join } from "node:path";
+import { pathToFileURL } from "node:url";
+import satori from "satori";
+import { Resvg } from "@resvg/resvg-js";
+import { encode } from "jpeg-js";
+import { loadEpisodes } from "../src/content/load.ts";
+import type { Episode } from "../src/content/parse.ts";
+import { cardTitle } from "../src/content/slug.ts";
+
+export { cardTitle };
+
+const W = 1200;
+const H = 630;
+const PANEL = W - H; // 570px of text to the right of the square artwork
+
+// Sampled from public/artwork.jpg: the wordmark navy and a cream pulled from
+// the top-left of the gradient.
+const NAVY = "#123F8C";
+const CREAM = "#FAF1E2";
+const MUTED = "#6A7BA0";
+
+const OUT = "dist/client/og";
+const JPEG_QUALITY = 82;
+
+export function runtime(seconds: number | undefined): string {
+  if (!seconds) return "";
+  const h = Math.floor(seconds / 3600);
+  const m = Math.round((seconds % 3600) / 60);
+  return h ? `${h}h ${m}m` : `${m}m`;
+}
+
+export function monthYear(date: string): string {
+  return new Date(`${date.slice(0, 10)}T00:00:00Z`).toLocaleDateString(
+    "en-US",
+    {
+      month: "long",
+      year: "numeric",
+      timeZone: "UTC",
+    },
+  );
+}
+
+type Node = { type: string; props: Record<string, unknown> };
+const el = (type: string, props: Record<string, unknown>): Node => ({
+  type,
+  props,
+});
+
+function card(artwork: string, title: string, meta: string[]): Node {
+  // Long titles get a smaller face so three lines still say something useful.
+  const size = title.length > 58 ? 34 : title.length > 34 ? 40 : 46;
+  return el("div", {
+    style: { display: "flex", width: W, height: H, backgroundColor: CREAM },
+    children: [
+      el("img", { src: artwork, width: H, height: H }),
+      el("div", {
+        style: {
+          display: "flex",
+          flexDirection: "column",
+          justifyContent: "center",
+          width: PANEL,
+          height: H,
+          padding: "56px 48px",
+        },
+        children: [
+          el("div", {
+            style: {
+              display: "-webkit-box",
+              "-webkit-box-orient": "vertical",
+              "-webkit-line-clamp": 3,
+              overflow: "hidden",
+              textOverflow: "ellipsis",
+              color: NAVY,
+              fontFamily: "Inter",
+              fontWeight: 700,
+              fontSize: size,
+              lineHeight: 1.2,
+              letterSpacing: "-0.02em",
+            },
+            children: title,
+          }),
+          el("div", {
+            style: {
+              display: "flex",
+              flexDirection: "column",
+              marginTop: 28,
+              color: MUTED,
+              fontFamily: "Inter",
+              fontWeight: 400,
+              fontSize: 21,
+              lineHeight: 1.45,
+            },
+            children: meta
+              .filter(Boolean)
+              .map((line) =>
+                el("div", { style: { display: "flex" }, children: line }),
+              ),
+          }),
+        ],
+      }),
+    ],
+  });
+}
+
+async function main() {
+  const root = process.cwd();
+  const [regular, bold, artworkBytes, episodes] = await Promise.all([
+    readFile(join(root, "assets/fonts/Inter-Regular.woff")),
+    readFile(join(root, "assets/fonts/Inter-Bold.woff")),
+    readFile(join(root, "public/artwork.jpg")),
+    loadEpisodes(),
+  ]);
+  const artwork = `data:image/jpeg;base64,${artworkBytes.toString("base64")}`;
+  const fonts = [
+    {
+      name: "Inter",
+      data: regular,
+      weight: 400 as const,
+      style: "normal" as const,
+    },
+    {
+      name: "Inter",
+      data: bold,
+      weight: 700 as const,
+      style: "normal" as const,
+    },
+  ];
+
+  await mkdir(join(root, OUT), { recursive: true });
+
+  const render = async (name: string, node: Node) => {
+    const svg = await satori(node as never, { width: W, height: H, fonts });
+    // Straight from resvg's RGBA buffer into a JPEG — the artwork's film grain
+    // defeats PNG's filters (~700 KB/card), and skipping asPng() avoids
+    // encoding a PNG we would only throw away.
+    const { pixels, width, height } = new Resvg(svg).render();
+    const jpg = encode({ data: pixels, width, height }, JPEG_QUALITY).data;
+    await writeFile(join(root, OUT, `${name}.jpg`), jpg);
+  };
+
+  const hosts = (ep: Episode) =>
+    ep.people
+      .map((p) => p.name)
+      .slice(0, 3)
+      .join(", ");
+
+  const started = Date.now();
+  for (const ep of episodes) {
+    await render(
+      ep.slug,
+      card(artwork, cardTitle(ep.title), [
+        monthYear(ep.date),
+        hosts(ep),
+        runtime(ep.duration),
+      ]),
+    );
+  }
+  await render(
+    "default",
+    card(artwork, "This Month in React", [
+      "Monthly React news, with full transcripts",
+    ]),
+  );
+  console.log(
+    `og: ${episodes.length + 1} cards in ${Date.now() - started}ms -> ${OUT}/`,
+  );
+}
+
+if (
+  process.argv[1] &&
+  import.meta.url === pathToFileURL(process.argv[1]).href
+) {
+  await main();
+}
diff --git a/scripts/publish-atproto.ts b/scripts/publish-atproto.ts
index 7672a2a..67c6e31 100644
--- a/scripts/publish-atproto.ts
+++ b/scripts/publish-atproto.ts
@@ -21,7 +21,8 @@ export function outlineToText(items: OutlineItem[], depth = 0): string {
   for (const item of items) {
     const indent = "  ".repeat(depth);
     lines.push(`${indent}${item.title}${item.url ? ` — ${item.url}` : ""}`);
-    if (item.children.length > 0) lines.push(outlineToText(item.children, depth + 1));
+    if (item.children.length > 0)
+      lines.push(outlineToText(item.children, depth + 1));
   }
   return lines.join("\n");
 }
@@ -30,7 +31,9 @@ export function outlineToText(items: OutlineItem[], depth = 0): string {
 export function bskyUrlToParts(
   url: string,
 ): { actor: string; rkey: string } | null {
-  const m = /^https:\/\/bsky\.app\/profile\/([^/]+)\/post\/([^/?#]+)/.exec(url.trim());
+  const m = /^https:\/\/bsky\.app\/profile\/([^/]+)\/post\/([^/?#]+)/.exec(
+    url.trim(),
+  );
   return m ? { actor: m[1], rkey: m[2] } : null;
 }
 
@@ -87,7 +90,8 @@ async function resolvePostRef(
   }
   const did = parts.actor.startsWith("did:")
     ? parts.actor
-    : (await agent.com.atproto.identity.resolveHandle({ handle: parts.actor })).data.did;
+    : (await agent.com.atproto.identity.resolveHandle({ handle: parts.actor }))
+        .data.did;
   const res = await agent.com.atproto.repo.getRecord({
     repo: did,
     collection: "app.bsky.feed.post",
@@ -129,7 +133,8 @@ async function main() {
     .filter((n) => n.endsWith(".md"))
     .filter((n) => !only || n === `${only}.md`)
     .sort();
-  if (only && names.length === 0) throw new Error(`no episode file for ${only}`);
+  if (only && names.length === 0)
+    throw new Error(`no episode file for ${only}`);
 
   for (const name of names) {
     const path = join(EPISODE_DIR, name);
diff --git a/scripts/publish-transcript.ts b/scripts/publish-transcript.ts
index 647d808..63b7b4c 100644
--- a/scripts/publish-transcript.ts
+++ b/scripts/publish-transcript.ts
@@ -52,10 +52,12 @@ export function descriptToCanonical(markdown: string): string {
 /**
  * A Descript export carries no `## ` headings, so a transcript written from
  * one straight would have no sections at all. Synthesize them from the
- * episode's existing outline, the same way scripts/migrate.ts does for the
- * legacy raw-Descript sources.
+ * episode's existing outline.
  */
-export function addOutlineHeadings(fileText: string, transcriptBody: string): string {
+export function addOutlineHeadings(
+  fileText: string,
+  transcriptBody: string,
+): string {
   const { outline } = parseEpisode(fileText, "");
   const items = outline
     .filter((item) => item.time)
@@ -65,9 +67,13 @@ export function addOutlineHeadings(fileText: string, transcriptBody: string): st
 }
 
 /** Swap everything below `# Transcript`, leaving front matter and outline alone. */
-export function replaceTranscript(fileText: string, transcriptBody: string): string {
+export function replaceTranscript(
+  fileText: string,
+  transcriptBody: string,
+): string {
   const marker = fileText.indexOf(`\n${TRANSCRIPT_MARKER}\n`);
-  if (marker === -1) throw new Error(`file has no "${TRANSCRIPT_MARKER}" heading`);
+  if (marker === -1)
+    throw new Error(`file has no "${TRANSCRIPT_MARKER}" heading`);
   const head = fileText.slice(0, marker + TRANSCRIPT_MARKER.length + 1);
   return `${head}\n${transcriptBody.trimEnd()}\n`;
 }
@@ -91,9 +97,21 @@ export async function fetchDescriptTranscript(
     }),
   });
   if (!res.ok) {
-    throw new Error(`Descript export failed: ${res.status} ${await res.text()}`);
+    throw new Error(
+      `Descript export failed: ${res.status} ${await res.text()}`,
+    );
+  }
+  const text = await res.text();
+  // A 200 with an empty or stub body would otherwise canonicalize to "" and
+  // wipe the committed transcript. The shortest real episode transcript runs
+  // to tens of kilobytes, so anything under 500 characters is a broken export,
+  // not a short episode.
+  if (text.trim().length < 500) {
+    throw new Error(
+      `Descript export returned only ${text.trim().length} characters; refusing to overwrite the transcript`,
+    );
   }
-  return res.text();
+  return text;
 }
 
 async function findTransistorEpisodeId(
@@ -104,7 +122,10 @@ async function findTransistorEpisodeId(
   for (let page = 1; page <= 20; page += 1) {
     const url = `${TRANSISTOR_BASE}/v1/episodes?show_id=${encodeURIComponent(showId)}&pagination[page]=${page}&pagination[per]=50`;
     const res = await fetch(url, { headers: { "x-api-key": apiKey } });
-    if (!res.ok) throw new Error(`Transistor list failed: ${res.status} ${await res.text()}`);
+    if (!res.ok)
+      throw new Error(
+        `Transistor list failed: ${res.status} ${await res.text()}`,
+      );
     const body = (await res.json()) as {
       data: { id: string; attributes: { share_url?: string } }[];
     };
@@ -130,7 +151,9 @@ export async function pushSrtToTranscript(
     body: JSON.stringify({ episode: { transcript_text: srt } }),
   });
   if (!res.ok) {
-    throw new Error(`Transistor update failed: ${res.status} ${await res.text()}`);
+    throw new Error(
+      `Transistor update failed: ${res.status} ${await res.text()}`,
+    );
   }
 }
 
@@ -157,7 +180,10 @@ async function main() {
   let fileText = readFileSync(path, "utf8");
 
   if (!skipDescript) {
-    const markdown = await fetchDescriptTranscript(projectId, requireEnv("DESCRIPT_TOKEN"));
+    const markdown = await fetchDescriptTranscript(
+      projectId,
+      requireEnv("DESCRIPT_TOKEN"),
+    );
     const body = addOutlineHeadings(fileText, descriptToCanonical(markdown));
     fileText = replaceTranscript(fileText, body);
     writeFileSync(path, fileText);
@@ -176,7 +202,9 @@ async function main() {
       requireEnv("TRANSISTOR_API_KEY"),
       requireEnv("TRANSISTOR_SHOW_ID"),
     );
-    console.log(`pushed ${srt.split("\n\n").length} SRT cues to Transistor episode ${episode.transistorId}`);
+    console.log(
+      `pushed ${srt.split("\n\n").length} SRT cues to Transistor episode ${episode.transistorId}`,
+    );
   }
 }
 
diff --git a/src/components/Document.tsx b/src/components/Document.tsx
index cb88695..5beef56 100644
--- a/src/components/Document.tsx
+++ b/src/components/Document.tsx
@@ -4,21 +4,66 @@ export const SITE_URL = (
   import.meta.env.VITE_SITE_URL || "https://thismonthinreact.com"
 ).replace(/\/$/, "");
 
+export const SITE_DESCRIPTION = "Monthly React news, with full transcripts.";
+
+/** The navy of the wordmark and of the generated OG cards. */
+export const THEME_COLOR = "#123F8C";
+
 export interface Og {
   title: string;
   description: string;
   url: string;
   /** "article" for episodes, "website" for everything else. */
   type?: "article" | "website";
+  /** Path under /og; the default card covers every non-episode page. */
+  image?: string;
+  imageAlt?: string;
+  /** Episodes only. */
+  publishedTime?: string;
+  audioUrl?: string;
 }
 
-/** Open Graph tags as router `head().meta` descriptors. No image: none exists yet. */
-export function ogMeta({ title, description, url, type = "website" }: Og) {
+/**
+ * Open Graph tags as router `head().meta` descriptors. Image URLs are absolute
+ * because relative ones do not unfurl in Discord or Slack.
+ */
+export function ogMeta({
+  title,
+  description,
+  url,
+  type = "website",
+  image = "default",
+  imageAlt,
+  publishedTime,
+  audioUrl,
+}: Og): { property?: string; name?: string; content: string }[] {
+  const imageUrl = `${SITE_URL}/og/${image}.jpg`;
   return [
     { property: "og:title", content: title },
     { property: "og:description", content: description },
     { property: "og:url", content: url },
     { property: "og:type", content: type },
+    { property: "og:locale", content: "en_US" },
+    { property: "og:image", content: imageUrl },
+    { property: "og:image:width", content: "1200" },
+    { property: "og:image:height", content: "630" },
+    { property: "og:image:type", content: "image/jpeg" },
+    { property: "og:image:alt", content: imageAlt ?? title },
+    ...(publishedTime
+      ? [{ property: "article:published_time", content: publishedTime }]
+      : []),
+    // Best-effort: standards-correct audio tags. Neither Discord nor Slack
+    // grants an inline player to an arbitrary site.
+    ...(audioUrl
+      ? [
+          { property: "og:audio", content: audioUrl },
+          { property: "og:audio:secure_url", content: audioUrl },
+          { property: "og:audio:type", content: "audio/mpeg" },
+        ]
+      : []),
+    { name: "twitter:title", content: title },
+    { name: "twitter:description", content: description },
+    { name: "twitter:image", content: imageUrl },
   ];
 }
 
@@ -27,7 +72,7 @@ export function OgTags(props: Og) {
   return (
     <>
       {ogMeta(props).map((m) => (
-        <meta key={m.property} property={m.property} content={m.content} />
+        <meta key={m.property ?? m.name} {...m} />
       ))}
     </>
   );
@@ -57,7 +102,9 @@ export function Document({
         <meta name="viewport" content="width=device-width, initial-scale=1" />
         <link rel="stylesheet" href="/styles.css" />
         <meta property="og:site_name" content={SITE_NAME} />
-        <meta name="twitter:card" content="summary" />
+        {/* summary_large_image is what makes Discord render the card full-width. */}
+        <meta name="twitter:card" content="summary_large_image" />
+        <meta name="theme-color" content={THEME_COLOR} />
         <link
           rel="alternate"
           type="application/rss+xml"
@@ -67,19 +114,22 @@ export function Document({
         {head}
       </head>
       <body className={bodyClass}>
+        <a className="skip-link" href="#main" data-pagefind-ignore="">
+          Skip to content
+        </a>
         <header className="site-header" data-pagefind-ignore="">
           <a className="site-name" href="/">
             {SITE_NAME}
           </a>
           <nav className="site-nav">
             <a href="/">Home</a>
-            <a href="/">Episodes</a>
+            <a href="/#archive">Episodes</a>
             <a href="/links">Links</a>
             <a href="/search">Search</a>
             <a href="/about">About</a>
           </nav>
         </header>
-        <main className={bodyClass ? `${bodyClass}-main` : undefined}>
+        <main id="main" className={bodyClass ? `${bodyClass}-main` : undefined}>
           {children}
         </main>
         <footer className="site-footer" data-pagefind-ignore="">
diff --git a/src/components/PagefindUI.tsx b/src/components/PagefindUI.tsx
index c5bdbca..ed9843c 100644
--- a/src/components/PagefindUI.tsx
+++ b/src/components/PagefindUI.tsx
@@ -21,7 +21,10 @@ export function PagefindUI() {
     script.src = "/pagefind/pagefind-ui.js";
     script.onload = () => {
       if (window.PagefindUI) {
-        new window.PagefindUI({ element: "#pagefind-ui", showSubResults: true });
+        new window.PagefindUI({
+          element: "#pagefind-ui",
+          showSubResults: true,
+        });
       }
     };
     document.head.appendChild(script);
diff --git a/src/components/Player.tsx b/src/components/Player.tsx
index 7a754a2..7025671 100644
--- a/src/components/Player.tsx
+++ b/src/components/Player.tsx
@@ -1,31 +1,45 @@
-'use client'
-import { useEffect, useRef } from 'react'
+"use client";
+import { useEffect, useRef } from "react";
 
 // Same [data-seconds] -> audio.currentTime seek convention as EpisodeBody's
 // SEEK_SCRIPT (deliberately duplicated: this is a hydrated component, that is
 // an inline script for the static episode document).
-export function Player({ audioUrl, title }: { audioUrl?: string; title: string }) {
-  const ref = useRef<HTMLAudioElement>(null)
+export function Player({
+  audioUrl,
+  title,
+}: {
+  audioUrl?: string;
+  title: string;
+}) {
+  const ref = useRef<HTMLAudioElement>(null);
 
   useEffect(() => {
     function onClick(event: MouseEvent) {
-      const target = (event.target as HTMLElement | null)?.closest<HTMLElement>('[data-seconds]')
-      const audio = ref.current
-      if (!target || !audio) return
-      const seconds = Number(target.dataset.seconds)
-      if (!Number.isFinite(seconds)) return
-      event.preventDefault()
-      audio.currentTime = seconds
-      void audio.play()
+      const target = (event.target as HTMLElement | null)?.closest<HTMLElement>(
+        "[data-seconds]",
+      );
+      const audio = ref.current;
+      if (!target || !audio) return;
+      const seconds = Number(target.dataset.seconds);
+      if (!Number.isFinite(seconds)) return;
+      event.preventDefault();
+      audio.currentTime = seconds;
+      void audio.play();
     }
-    document.addEventListener('click', onClick)
-    return () => document.removeEventListener('click', onClick)
-  }, [])
+    document.addEventListener("click", onClick);
+    return () => document.removeEventListener("click", onClick);
+  }, []);
 
-  if (!audioUrl) return null
+  if (!audioUrl) return null;
   return (
     <div className="player" data-pagefind-ignore="">
-      <audio ref={ref} controls preload="none" src={audioUrl} aria-label={`Play ${title}`} />
+      <audio
+        ref={ref}
+        controls
+        preload="none"
+        src={audioUrl}
+        aria-label={`Play ${title}`}
+      />
     </div>
-  )
+  );
 }
diff --git a/src/content/feed.ts b/src/content/feed.ts
index 255db67..b9edad6 100644
--- a/src/content/feed.ts
+++ b/src/content/feed.ts
@@ -18,7 +18,9 @@ export function outlineToHtml(items: OutlineItem[]): string {
   if (items.length === 0) return "";
   const lis = items.map((item) => {
     const title = escapeHtml(item.title);
-    const label = item.url ? `<a href="${escapeHtml(item.url)}">${title}</a>` : title;
+    const label = item.url
+      ? `<a href="${escapeHtml(item.url)}">${title}</a>`
+      : title;
     return `<li>${label}${outlineToHtml(item.children)}</li>`;
   });
   return `<ul>${lis.join("")}</ul>`;
@@ -45,7 +47,11 @@ export function outlineToChapters(
   return out.sort((a, b) => a.startTime - b.startTime);
 }
 
-export function renderFeed(episodes: Episode[], siteName: string, siteUrl: string): string {
+export function renderFeed(
+  episodes: Episode[],
+  siteName: string,
+  siteUrl: string,
+): string {
   const items = episodes
     .map((episode) => {
       const url = `${siteUrl}/episodes/${episode.slug}`;
diff --git a/src/content/headings.ts b/src/content/headings.ts
index 21f9220..6491cf0 100644
--- a/src/content/headings.ts
+++ b/src/content/headings.ts
@@ -17,25 +17,27 @@ function paragraphSeconds(block: string): number | undefined {
  * Synthesize `## <title>` section headings for a canonical transcript that has
  * none, from timed outline items. Each item's heading is inserted directly
  * before the first paragraph at or past that item's time. An item with no such
- * paragraph, or one that would land on the same paragraph as the item before
- * it (no content between them), is skipped rather than opening an empty
+ * paragraph, or one that would land on a paragraph another item
+ * already claimed (no content between them), is skipped rather than opening an empty
  * section.
  *
- * Shared by scripts/migrate.ts (legacy sources with a raw Descript body) and
- * scripts/publish-transcript.ts (a fresh Descript export, which also carries
- * no headings of its own).
+ * Used by scripts/publish-transcript.ts, since a Descript export carries no
+ * headings of its own.
  */
-export function insertHeadings(paragraphs: string[], items: TimedItem[]): string[] {
+export function insertHeadings(
+  paragraphs: string[],
+  items: TimedItem[],
+): string[] {
   const times = paragraphs.map(paragraphSeconds);
   const headingAt = new Map<number, string>();
-  let lastTarget = -1;
 
   for (const item of items) {
     const sec = timestampToSeconds(item.time);
     const idx = times.findIndex((t) => t !== undefined && t >= sec);
-    if (idx === -1 || idx === lastTarget) continue;
+    // First item to claim a paragraph keeps it: outlines drift out of order, and
+    // overwriting the entry would silently drop the heading already placed there.
+    if (idx === -1 || headingAt.has(idx)) continue;
     headingAt.set(idx, `## ${item.title}`);
-    lastTarget = idx;
   }
 
   const out: string[] = [];
diff --git a/src/content/jsonld.ts b/src/content/jsonld.ts
new file mode 100644
index 0000000..4e0a263
--- /dev/null
+++ b/src/content/jsonld.ts
@@ -0,0 +1,8 @@
+/**
+ * Serialize structured data for a <script type="application/ld+json"> block.
+ * `<` becomes `\u003c` — still valid JSON, but a `</script>` inside a title or
+ * description can no longer close the block.
+ */
+export function jsonLd(data: unknown): string {
+  return JSON.stringify(data).replace(/</g, "\\u003c");
+}
diff --git a/src/content/links.ts b/src/content/links.ts
index da169c7..3596c1d 100644
--- a/src/content/links.ts
+++ b/src/content/links.ts
@@ -20,14 +20,15 @@ function walk(items: OutlineItem[], episode: Episode, out: LinkEntry[]): void {
         // Not an absolute URL (e.g. an in-page anchor like "#section") — no
         // external link to index.
       }
-      if (host) out.push({
-        text: item.title,
-        url: item.url,
-        host,
-        episodeSlug: episode.slug,
-        episodeTitle: episode.title,
-        time: toSeconds(item.time),
-      });
+      if (host)
+        out.push({
+          text: item.title,
+          url: item.url,
+          host,
+          episodeSlug: episode.slug,
+          episodeTitle: episode.title,
+          time: toSeconds(item.time),
+        });
     }
     walk(item.children, episode, out);
   }
@@ -39,7 +40,9 @@ export function buildLinkIndex(episodes: Episode[]): LinkEntry[] {
   return out;
 }
 
-export function groupByHost(entries: LinkEntry[]): { host: string; entries: LinkEntry[] }[] {
+export function groupByHost(
+  entries: LinkEntry[],
+): { host: string; entries: LinkEntry[] }[] {
   const map = new Map<string, LinkEntry[]>();
   for (const entry of entries) {
     const list = map.get(entry.host);
@@ -48,5 +51,8 @@ export function groupByHost(entries: LinkEntry[]): { host: string; entries: Link
   }
   return [...map]
     .map(([host, list]) => ({ host, entries: list }))
-    .sort((a, b) => b.entries.length - a.entries.length || a.host.localeCompare(b.host));
+    .sort(
+      (a, b) =>
+        b.entries.length - a.entries.length || a.host.localeCompare(b.host),
+    );
 }
diff --git a/src/content/parse.ts b/src/content/parse.ts
index 790ef66..eabb9ff 100644
--- a/src/content/parse.ts
+++ b/src/content/parse.ts
@@ -228,7 +228,10 @@ function flattenOutline(items: OutlineItem[]): OutlineItem[] {
  * drifting backward onto an unrelated earlier section (e.g. "Intro" at
  * 00:00:00, which sits within tolerance of 00:01:51 by raw distance alone).
  */
-function reconcileOutlineAnchors(roots: OutlineItem[], sections: Section[]): void {
+function reconcileOutlineAnchors(
+  roots: OutlineItem[],
+  sections: Section[],
+): void {
   const sectionByAnchor = new Map(sections.map((s) => [s.anchor, s]));
   const items = flattenOutline(roots);
   const claimed = new Set<Section>();
@@ -262,7 +265,8 @@ function reconcileOutlineAnchors(roots: OutlineItem[], sections: Section[]): voi
   for (const item of items) {
     const resolved = resolvedFor.get(item);
     if (resolved) {
-      if (resolved.time) floor = Math.max(floor, timestampToSeconds(resolved.time));
+      if (resolved.time)
+        floor = Math.max(floor, timestampToSeconds(resolved.time));
       continue;
     }
     if (!item.time) continue;
@@ -289,20 +293,36 @@ function reconcileOutlineAnchors(roots: OutlineItem[], sections: Section[]): voi
 
 export function parseEpisode(text: string, epSlug: string): Episode {
   const { frontMatter, body } = splitFile(text);
-  const marker = body.indexOf(`\n${TRANSCRIPT_MARKER}\n`);
+  // The marker starts a line: either at position 0 (a file with no outline,
+  // whose body begins with it) or just after a newline.
+  const afterNewline = body.indexOf(`\n${TRANSCRIPT_MARKER}\n`);
+  const marker = body.startsWith(`${TRANSCRIPT_MARKER}\n`)
+    ? 0
+    : afterNewline === -1
+      ? -1
+      : afterNewline + 1;
   const outlineRegion = marker === -1 ? body : body.slice(0, marker);
   const transcriptRegion =
-    marker === -1 ? "" : body.slice(marker + TRANSCRIPT_MARKER.length + 2);
+    marker === -1 ? "" : body.slice(marker + TRANSCRIPT_MARKER.length + 1);
 
   const outline = parseOutline(outlineRegion);
   const sections = parseSections(transcriptRegion);
   reconcileOutlineAnchors(outline, sections);
 
   const fm = frontMatter as Record<string, any>;
+  // A missing or unparseable date silently becomes `<pubDate>Invalid Date</pubDate>`
+  // in the feed and throws mid-loop in publish-atproto, after records are already
+  // written. Fail here, naming the episode, before any of that runs.
+  const date = String(fm.date ?? "");
+  if (Number.isNaN(Date.parse(date))) {
+    throw new Error(
+      `episode ${epSlug || "(unnamed)"}: missing or unparseable front matter date: ${JSON.stringify(fm.date ?? null)}`,
+    );
+  }
   return {
     slug: epSlug,
     title: String(fm.title ?? ""),
-    date: String(fm.date ?? ""),
+    date,
     description: String(fm.description ?? ""),
     time: fm.time,
     location: fm.location,
diff --git a/src/content/slug.ts b/src/content/slug.ts
index 93a9d0c..c375ef8 100644
--- a/src/content/slug.ts
+++ b/src/content/slug.ts
@@ -51,3 +51,18 @@ export function secondsToTimestamp(total: number): string {
 export function normalizeTime(raw: string): string {
   return secondsToTimestamp(timestampToSeconds(raw.trim()));
 }
+
+/**
+ * Episode titles carry boilerplate — a "TMiR 2026-07: " prefix on recent ones,
+ * a "This Month in React, July 2024: " prefix on older ones, and a trailing
+ * "(March 2023)" on the oldest — that duplicates the site name and the date
+ * shown alongside.
+ */
+export function cardTitle(title: string): string {
+  const stripped = title
+    .replace(/^TMiR\s+\d{4}-\d{2}:\s*/, "")
+    .replace(/^This Month in React,[^:]*:\s*/, "")
+    .replace(/\s*\([A-Z][a-z]+ \d{4}\)$/, "")
+    .trim();
+  return stripped || title;
+}
diff --git a/src/content/srt.ts b/src/content/srt.ts
index 22e7a68..fa43ae0 100644
--- a/src/content/srt.ts
+++ b/src/content/srt.ts
@@ -2,6 +2,11 @@ import type { Episode } from "./parse.ts";
 import { secondsToTimestamp, timestampToSeconds } from "./slug.ts";
 
 const LAST_CUE_SECONDS = 5;
+// Transcript timestamps are whole seconds and two segments can share one (or,
+// where an outline drifts, run backwards), which would emit a zero-length or
+// reversed cue that players and Transistor reject. One second is the smallest
+// duration this format can express, so it's the floor.
+const MIN_CUE_SECONDS = 1;
 
 function cueTime(seconds: number): string {
   return `${secondsToTimestamp(seconds)},000`;
@@ -14,12 +19,17 @@ export function toSrt(episode: Episode): string {
     .filter((segment) => segment.time)
     .map((segment) => ({
       start: timestampToSeconds(segment.time!),
-      text: segment.speaker ? `${segment.speaker}: ${segment.text}` : segment.text,
+      text: segment.speaker
+        ? `${segment.speaker}: ${segment.text}`
+        : segment.text,
     }));
 
   return cues
     .map((cue, i) => {
-      const end = cues[i + 1]?.start ?? cue.start + LAST_CUE_SECONDS;
+      const end = Math.max(
+        cue.start + MIN_CUE_SECONDS,
+        cues[i + 1]?.start ?? cue.start + LAST_CUE_SECONDS,
+      );
       return `${i + 1}\n${cueTime(cue.start)} --> ${cueTime(end)}\n${cue.text}\n`;
     })
     .join("\n");
diff --git a/src/content/time.ts b/src/content/time.ts
index e49503e..051b310 100644
--- a/src/content/time.ts
+++ b/src/content/time.ts
@@ -6,10 +6,15 @@ export function toSeconds(
   if (value === undefined) return undefined;
   if (typeof value === "number")
     return Number.isFinite(value) ? value : undefined;
-  const parts = value.split(":").map(Number);
-  if (parts.length === 0 || parts.some((n) => !Number.isFinite(n)))
-    return undefined;
-  return parts.reduce((total, part) => total * 60 + part, 0);
+  // `""` splits to `[""]` and `Number("")` is 0, so an empty or blank field
+  // would otherwise read as a valid 0 seconds. At most h:m:s, too.
+  const parts = value.split(":");
+  if (parts.length > 3) return undefined;
+  const numbers = parts.map((part) =>
+    part.trim() === "" ? NaN : Number(part),
+  );
+  if (numbers.some((n) => !Number.isFinite(n))) return undefined;
+  return numbers.reduce((total, part) => total * 60 + part, 0);
 }
 
 export function hms(seconds: number): string {
diff --git a/src/routes/about.tsx b/src/routes/about.tsx
index 22ccb07..b99e104 100644
--- a/src/routes/about.tsx
+++ b/src/routes/about.tsx
@@ -32,7 +32,10 @@ const getAbout = createServerFn().handler(async () => {
   const byName = new Map<string, AboutPerson>();
   for (const episode of episodes) {
     for (const person of episode.people) {
-      const entry = byName.get(person.name) ?? { name: person.name, role: "Guest" };
+      const entry = byName.get(person.name) ?? {
+        name: person.name,
+        role: "Guest",
+      };
       entry.href ||= person.href;
       entry.img ||= person.img;
       byName.set(person.name, entry);
@@ -74,7 +77,13 @@ function PersonList({ people }: { people: AboutPerson[] }) {
       {people.map((person) => (
         <li key={person.name}>
           {person.img && (
-            <img src={person.img} alt={person.name} width="48" height="48" loading="lazy" />
+            <img
+              src={person.img}
+              alt={person.name}
+              width="48"
+              height="48"
+              loading="lazy"
+            />
           )}
           {person.href ? (
             <a href={person.href} rel="noreferrer">
@@ -122,7 +131,10 @@ function About() {
         <h2>Subscribe</h2>
         <ul>
           <li>
-            <a href="https://open.spotify.com/show/4g3Le83YfsMeI8Fq3cpPeH" rel="noreferrer">
+            <a
+              href="https://open.spotify.com/show/4g3Le83YfsMeI8Fq3cpPeH"
+              rel="noreferrer"
+            >
               Spotify
             </a>
           </li>
@@ -168,7 +180,13 @@ function About() {
             method="post"
           >
             <label htmlFor="bd-email">Email</label>
-            <input id="bd-email" type="email" name="email" required autoComplete="email" />
+            <input
+              id="bd-email"
+              type="email"
+              name="email"
+              required
+              autoComplete="email"
+            />
             <button type="submit">Subscribe</button>
           </form>
         </section>
diff --git a/src/routes/episodes.$slug.tsx b/src/routes/episodes.$slug.tsx
index c42da5c..5386b08 100644
--- a/src/routes/episodes.$slug.tsx
+++ b/src/routes/episodes.$slug.tsx
@@ -1,6 +1,12 @@
 import { createFileRoute } from "@tanstack/react-router";
 import { renderToStaticMarkup } from "react-dom/server";
-import { Document, OgTags, SITE_NAME, SITE_URL } from "../components/Document";
+import {
+  Document,
+  OgTags,
+  SITE_DESCRIPTION,
+  SITE_NAME,
+  SITE_URL,
+} from "../components/Document";
 import {
   COMMENTS_SCRIPT,
   EpisodeBody,
@@ -8,6 +14,9 @@ import {
   SEEK_SCRIPT,
 } from "../components/EpisodeBody";
 import { bskyPostToAtUri } from "../content/atproto.ts";
+import { jsonLd } from "../content/jsonld.ts";
+import { cardTitle } from "../content/slug.ts";
+import { isoDuration } from "../content/time.ts";
 
 const ATPROTO_DID = import.meta.env.VITE_ATPROTO_DID;
 
@@ -24,25 +33,65 @@ export const Route = createFileRoute("/episodes/$slug")({
           ? bskyPostToAtUri(episode.bskyPostUrl, ATPROTO_DID)
           : undefined;
 
+        // episode.title already carries the site name or a "TMiR <yyyy-mm>: "
+        // prefix, so strip it before appending the site name.
+        const name = cardTitle(episode.title);
+        const title = `${name} — ${SITE_NAME}`;
+        const description = episode.description || SITE_DESCRIPTION;
+        const url = `${SITE_URL}/episodes/${episode.slug}`;
+        const published = new Date(
+          `${episode.date.slice(0, 10)}T00:00:00Z`,
+        ).toISOString();
+
+        const structuredData = jsonLd({
+          "@context": "https://schema.org",
+          "@type": "PodcastEpisode",
+          name,
+          description,
+          url,
+          datePublished: published,
+          image: `${SITE_URL}/og/${episode.slug}.jpg`,
+          partOfSeries: {
+            "@type": "PodcastSeries",
+            name: SITE_NAME,
+            url: `${SITE_URL}/`,
+          },
+          ...(episode.audioUrl
+            ? {
+                associatedMedia: {
+                  "@type": "AudioObject",
+                  contentUrl: episode.audioUrl,
+                  ...(episode.duration !== undefined
+                    ? { duration: isoDuration(episode.duration) }
+                    : {}),
+                },
+              }
+            : {}),
+        });
+
         const html = renderToStaticMarkup(
           <Document
             head={
               <>
-                <title>{`${episode.title} — ${SITE_NAME}`}</title>
-                <meta name="description" content={episode.description} />
-                <link
-                  rel="canonical"
-                  href={`${SITE_URL}/episodes/${episode.slug}`}
-                />
+                <title>{title}</title>
+                <meta name="description" content={description} />
+                <link rel="canonical" href={url} />
                 <OgTags
-                  title={`${episode.title} — ${SITE_NAME}`}
-                  description={episode.description}
-                  url={`${SITE_URL}/episodes/${episode.slug}`}
+                  title={title}
+                  description={description}
+                  url={url}
                   type="article"
+                  image={episode.slug}
+                  publishedTime={published}
+                  audioUrl={episode.audioUrl}
                 />
                 {episode.atUri && (
                   <link rel="site.standard.document" href={episode.atUri} />
                 )}
+                <script
+                  type="application/ld+json"
+                  dangerouslySetInnerHTML={{ __html: structuredData }}
+                />
               </>
             }
             bodyClass="episode-page"
@@ -59,7 +108,7 @@ export const Route = createFileRoute("/episodes/$slug")({
             }
           >
             <header className="episode-header">
-              <h1 title={episode.title}>{episode.title}</h1>
+              <h1>{episode.title}</h1>
               {episode.audioUrl && (
                 <div className="player" data-pagefind-ignore="">
                   <audio controls preload="none" src={episode.audioUrl} />
diff --git a/src/routes/index.tsx b/src/routes/index.tsx
index 348d393..2798360 100644
--- a/src/routes/index.tsx
+++ b/src/routes/index.tsx
@@ -3,6 +3,7 @@ import { createServerFn } from "@tanstack/react-start";
 import { Outline } from "../components/EpisodeBody";
 import { Player } from "../components/Player";
 import { SITE_NAME, SITE_URL, ogMeta } from "../components/Document";
+import { jsonLd } from "../content/jsonld.ts";
 import type { OutlineItem } from "../content/parse.ts";
 
 const HOME_DESCRIPTION = `Every episode of ${SITE_NAME}, with transcripts and links.`;
@@ -50,6 +51,20 @@ function Home() {
   const { latest, archive } = Route.useLoaderData();
   return (
     <>
+      <script
+        type="application/ld+json"
+        dangerouslySetInnerHTML={{
+          __html: jsonLd({
+            "@context": "https://schema.org",
+            "@type": "PodcastSeries",
+            name: SITE_NAME,
+            description: HOME_DESCRIPTION,
+            url: `${SITE_URL}/`,
+            image: `${SITE_URL}/artwork.jpg`,
+            webFeed: `${SITE_URL}/feed.xml`,
+          }),
+        }}
+      />
       {latest && (
         <section className="latest">
           <h1>
@@ -59,7 +74,7 @@ function Home() {
           <Outline items={latest.outline} />
         </section>
       )}
-      <section className="archive">
+      <section className="archive" id="archive">
         <h2>Archive</h2>
         <ul>
           {archive.map((e) => (
diff --git a/src/routes/links.tsx b/src/routes/links.tsx
index 288a537..0208553 100644
--- a/src/routes/links.tsx
+++ b/src/routes/links.tsx
@@ -23,9 +23,8 @@ export const Route = createFileRoute("/links")({
     handlers: {
       GET: async () => {
         const { loadEpisodes } = await import("../content/load.ts");
-        const { buildLinkIndex, groupByHost } = await import(
-          "../content/links.ts"
-        );
+        const { buildLinkIndex, groupByHost } =
+          await import("../content/links.ts");
         const { hms } = await import("../content/time.ts");
         const groups = groupByHost(buildLinkIndex(await loadEpisodes()));
 
@@ -62,7 +61,8 @@ export const Route = createFileRoute("/links")({
             {groups.map((group) => (
               <section className="link-group" key={group.host}>
                 <h2>
-                  {group.host} <span className="count">{group.entries.length}</span>
+                  {group.host}{" "}
+                  <span className="count">{group.entries.length}</span>
                 </h2>
                 <ul>
                   {group.entries.map((entry, i) => (
@@ -74,7 +74,10 @@ export const Route = createFileRoute("/links")({
                       <a href={entry.url} rel="noreferrer">
                         {entry.text}
                       </a>
-                      <a className="from" href={`/episodes/${entry.episodeSlug}`}>
+                      <a
+                        className="from"
+                        href={`/episodes/${entry.episodeSlug}`}
+                      >
                         {entry.episodeTitle}
                         {entry.time !== undefined && (
                           <span className="ts">{hms(entry.time)}</span>
diff --git a/src/routes/search.tsx b/src/routes/search.tsx
index c4e4ed4..7ff7414 100644
--- a/src/routes/search.tsx
+++ b/src/routes/search.tsx
@@ -24,8 +24,8 @@ function Search() {
       <h1>Search transcripts</h1>
       <noscript>
         <p>
-          Search needs JavaScript. Every transcript is on its episode page, and your
-          browser&rsquo;s find-in-page works there.
+          Search needs JavaScript. Every transcript is on its episode page, and
+          your browser&rsquo;s find-in-page works there.
         </p>
       </noscript>
       <PagefindUI />
diff --git a/src/routes/sitemap[.]xml.ts b/src/routes/sitemap[.]xml.ts
new file mode 100644
index 0000000..d53c28e
--- /dev/null
+++ b/src/routes/sitemap[.]xml.ts
@@ -0,0 +1,33 @@
+import { createFileRoute } from "@tanstack/react-router";
+import { SITE_URL } from "../components/Document";
+
+export const Route = createFileRoute("/sitemap.xml")({
+  server: {
+    handlers: {
+      GET: async () => {
+        const { loadEpisodes } = await import("../content/load.ts");
+        const episodes = await loadEpisodes();
+        // /search is deliberately absent: it carries its own noindex.
+        const urls: { path: string; lastmod?: string }[] = [
+          ...["/", "/about", "/links"].map((path) => ({ path })),
+          ...episodes.map((e) => ({
+            path: `/episodes/${e.slug}`,
+            lastmod: e.date.slice(0, 10),
+          })),
+        ];
+        const xml = [
+          '<?xml version="1.0" encoding="UTF-8"?>',
+          '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">',
+          ...urls.map(
+            ({ path, lastmod }) =>
+              `<url><loc>${SITE_URL}${path}</loc>${lastmod ? `<lastmod>${lastmod}</lastmod>` : ""}</url>`,
+          ),
+          "</urlset>",
+        ].join("");
+        return new Response(xml, {
+          headers: { "content-type": "application/xml; charset=utf-8" },
+        });
+      },
+    },
+  },
+});
diff --git a/tests/feed.test.ts b/tests/feed.test.ts
index ed51850..d83caaa 100644
--- a/tests/feed.test.ts
+++ b/tests/feed.test.ts
@@ -2,22 +2,25 @@ import { test } from "node:test";
 import assert from "node:assert/strict";
 
 process.env.TMIR_CONTENT_DIR = "tests/fixtures/episodes";
-const { cdata, outlineToHtml, outlineToChapters, renderFeed } = await import(
-  "../src/content/feed.ts"
-);
+const { cdata, outlineToHtml, outlineToChapters, renderFeed } =
+  await import("../src/content/feed.ts");
 const { loadEpisodes } = await import("../src/content/load.ts");
 
 test("outlineToHtml nests lists and links out", async () => {
   const [episode] = await loadEpisodes();
   const html = outlineToHtml(episode.outline);
   assert.match(html, /^<ul>/);
-  assert.ok(html.includes('<a href="https://react.dev/compiler">React Compiler</a>'));
+  assert.ok(
+    html.includes('<a href="https://react.dev/compiler">React Compiler</a>'),
+  );
   assert.ok(html.includes("<ul><li>"), "nested child list present");
   assert.ok(html.includes("No link here"));
 });
 
 test("outlineToHtml escapes markup in titles", () => {
-  const html = outlineToHtml([{ title: "a & b <script>", anchor: "a", children: [] }]);
+  const html = outlineToHtml([
+    { title: "a & b <script>", anchor: "a", children: [] },
+  ]);
   assert.ok(html.includes("a &amp; b &lt;script&gt;"));
   assert.ok(!html.includes("<script>"));
 });
@@ -48,7 +51,10 @@ test("renderFeed produces one item per episode with the outline as body", async
 test("enclosure carries no length, duration stays on itunes:duration", async () => {
   const xml = renderFeed(await loadEpisodes(), "TMiR", "https://example.com");
   assert.ok(xml.includes("<enclosure "), "enclosure present");
-  assert.ok(!/<enclosure[^>]*length=/.test(xml), "no length attribute (bytes unknown)");
+  assert.ok(
+    !/<enclosure[^>]*length=/.test(xml),
+    "no length attribute (bytes unknown)",
+  );
   assert.match(xml, /<itunes:duration>\d+<\/itunes:duration>/);
 });
 
diff --git a/tests/fixtures/legacy-2023-10.md b/tests/fixtures/legacy-2023-10.md
deleted file mode 100644
index c505ccf..0000000
--- a/tests/fixtures/legacy-2023-10.md
+++ /dev/null
@@ -1,21 +0,0 @@
----
-title: This Month in React (October 2023)
-date: 2023-10-25
-time: 10am PT / 5PM GMT
-location: Main Stage on Reactiflux
-description: "Join Carl Vitullo, and Mark Erikson as we break down This Month in React."
-people: "[Carl Vitullo](https://twitter.com/vcarl_) and [Mark Erikson](https://twitter.com/acemarke)"
----
-
-- [00:00] Introduction
-- [00:00] Also zero, lands on the same paragraph as Introduction
-- [01:00] Second topic
-- [10:00] Never happens, no paragraph reaches this time
-
-**Carl Vitullo:** Everyone, thank you for joining us. [00:00]
-
-Some news before we get going. [00:30]
-
-**Carl Vitullo:** On to the second topic now. [01:05]
-
-Wrapping up the second topic. [01:45]
diff --git a/tests/fixtures/legacy-2023-march.md b/tests/fixtures/legacy-2023-march.md
deleted file mode 100644
index faf7db1..0000000
--- a/tests/fixtures/legacy-2023-march.md
+++ /dev/null
@@ -1,24 +0,0 @@
----
-title: This Month in React (March 2023)
-date: 2023-03-29
-time: 10am PT / 5PM GMT
-location: Q&A Channel on Reactiflux
-description: "Join Carl Vitullo, Mark Erikson, and Matt Pocock as we break down This Month in React."
-people: "[Carl Vitullo](https://twitter.com/vcarl_), [Mark Erikson](https://twitter.com/acemarke), and [Matt Pocock](https://twitter.com/mattpocockuk)"
----
-
-<iframe src="https://podcasters.spotify.com/pod/show/reactiflux/embed/episodes/This-Month-In-React--March-2023-e21n2om" height="102px" width="400px" frameborder="0" scrolling="no"></iframe>
-
-**Carl Vitullo:** Thanks everyone for joining us for this month in React. [00:16]
-
-**Mark Erikson:** Sure. Hi, I'm Mark Erickson. I maintain Redux. [00:53]
-
-## Retire SPA and MPA?
-
-**Carl Vitullo:** So I'll kick us off here. Dan Abramov tweeted about retiring terms. [02:10]
-
-Yeah. What do you guys think about that? [02:38]
-
-## Create React App no longer officially recommended
-
-**Mark Erikson:** There definitely has been a shift in the discussion. [03:07]
diff --git a/tests/fixtures/legacy-2024-03.md b/tests/fixtures/legacy-2024-03.md
deleted file mode 100644
index a1fe1ce..0000000
--- a/tests/fixtures/legacy-2024-03.md
+++ /dev/null
@@ -1,33 +0,0 @@
----
-title: "This Month in React, March 2024: React canary is 19"
-date: 2024-03-27
-time: 10am PT / 5PM GMT
-location: Main Stage on Reactiflux
-description: "Join Carl Vitullo and Mark Erikson as we break down This Month in React."
-people: "[Carl Vitullo](https://twitter.com/vcarl_) and [Mark Erikson](https://twitter.com/acemarke)"
----
-
-- [Spotify](https://open.spotify.com/show/4g3Le83YfsMeI8Fq3cpPeH)
-- [Apple Podcasts](https://podcasts.apple.com/us/podcast/reactiflux-events/id1661733526)
-- [RSS](https://anchor.fm/s/a1f8a59c/podcast/rss)
-
-<iframe src="https://podcasters.spotify.com/pod/show/reactiflux/embed/episodes/TMiR-2024-03-e2hsspd" height="102px" width="400px" frameborder="0" scrolling="no"></iframe>
-
-- Quick Hits
-  - [[00:39](#layoffs-seems-better)] [Layoffs.fyi](https://layoffs.fyi/)
-  - [[01:20](#new-releases)] New releases
-    - [[06:55](#astro-45-astrodb)] [Astro 4.5](https://astro.build/blog/astro-450/)
-
-[00:22] **Carl Vitullo:** We got a lot of good stuff to discuss and digest here.
-
-## Layoffs (seems better!)
-
-[00:39] **Carl Vitullo:** I'm going to start off with some layoffs news.
-
-[00:58] This month has about 7, 200 laid off so far.
-
-## New releases
-
-[01:20] On to some new releases. We have a pretty big list this month.
-
-[01:23] **Carl Vitullo:** We've got some news from React Native.
diff --git a/tests/fixtures/legacy-2025-01.md b/tests/fixtures/legacy-2025-01.md
deleted file mode 100644
index 06b4e11..0000000
--- a/tests/fixtures/legacy-2025-01.md
+++ /dev/null
@@ -1,34 +0,0 @@
----
-title: "TMiR 2025-01: Movement on CRA, Redwood.js dead?"
-date: 2025-01-29
-time: 10am PT / 5PM GMT
-location: Main Stage on Reactiflux
-description: "Join Carl, Mark, and Mo as we break down This Month in React."
-people: "[Carl](https://twitter.com/vcarl_), [Mark](https://twitter.com/acemarke), and [Mo](https://twitter.com/mo__javad)"
----
-
-<style>
-iframe {
-  width: 100%;
-  height: 15rem;
-}
-html li p {
-  margin-bottom: 0;
-}
-</style>
-
-<iframe src="https://creators.spotify.com/pod/show/reactiflux/embed/episodes/TMiR-2025-01-e2u8ofd" height="102px" width="400px" frameborder="0" scrolling="no"></iframe>
-
-- [[01:29](#job-market-fred-data-layoffsfyi)] Job market: [FRED data](https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE), [Layoffs.fyi](https://layoffs.fyi/)
-- [[04:17](#new-releases)] New releases
-  - [[04:26](#react-query-563)] [React query 5.63](https://bsky.app/profile/tkdodo.eu/post/3lfaeteulds2i)
-
-**Mark:** am making the news. [01:26]
-
-**Carl:** making the news and then reporting on it. Cool.
-
-## Job market: [FRED data](https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE), [Layoffs.fyi](https://layoffs.fyi/)
-
-**Carl:** Yeah, let's start off briefly, overview of the job market. [01:50]
-
-But conferences still don't have a ton of queued up ones.
diff --git a/tests/fixtures/legacy-2025-06.md b/tests/fixtures/legacy-2025-06.md
deleted file mode 100644
index 9ed488f..0000000
--- a/tests/fixtures/legacy-2025-06.md
+++ /dev/null
@@ -1,37 +0,0 @@
----
-title: "TMiR 2025-06: kinda quiet tbh. ES2025 finalized, new Safari"
-date: 2025-06-30
-time: 10am PT / 5PM GMT
-location: Main Stage on Reactiflux
-description: "Join Carl, Mark, and Mo as we break down This Month in React."
-people: "[Carl](https://twitter.com/vcarl_), [Mark](https://twitter.com/acemarke), and [Mo](https://twitter.com/mo__javad)"
----
-
-<style>
-iframe {
-  position: sticky;
-  top: 0;
-}
-</style>
-
-<iframe width="100%" height="180" frameborder="no" scrolling="no" seamless="" src="https://share.transistor.fm/e/feba4de9"></iframe>
-
-We had some problems with transcript generation this month, downstream of the audio recording issues. Sorry! Fixed for July.
-
-- Job market [FRED data](https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE), [Layoffs.fyi](https://layoffs.fyi/)
-- New releases
-  - [Recharts v3.0](https://github.com/recharts/recharts/releases/tag/v3.0.0)
-
-# This Month in React: June, 2025
-
-[00:00] **1-vcarl:** Hi, Carl here. Just a quick note before we get into this episode.
-
-[00:51] **1-vcarl:** Endlessly!
-
-## [00:52] job market
-
-[00:52] **1-vcarl:** Start off with a little bit of job market news.
-
-## [02:19] New Releases
-
-[02:19] **1-vcarl:** Yeah. New releases.
diff --git a/tests/fixtures/legacy-2025-09.md b/tests/fixtures/legacy-2025-09.md
deleted file mode 100644
index f42095a..0000000
--- a/tests/fixtures/legacy-2025-09.md
+++ /dev/null
@@ -1,30 +0,0 @@
----
-title: "TMiR 2025-09: React 19.2 on the horizon; npm is still getting compromised"
-date: 2025-09-27
-time: 10am PT / 5PM GMT
-location: Main Stage on Reactiflux
-description: "Join Carl, Mark, and Mo as we break down This Month in React."
-people: "[Carl](https://twitter.com/vcarl_), [Mark](https://twitter.com/acemarke), and [Mo](https://twitter.com/mo__javad)"
----
-
-<style>
-iframe {
-  position: sticky;
-  top: 0;
-}
-</style>
-
-- [[00:52](#new-releases)] New releases
-  - [[00:54](#tanstack-start-10-rc)] [TanStack Start 1.0 RC!](https://tanstack.com/blog/announcing-tanstack-start-v1)
-
-Carl: Alright. Hello everyone. Thank you for joining us for the September edition of this Month In React. [00:00]
-
-Mark: Hi, I'm Mark. My day job is working at Replay.io. [00:23]
-
-Mo: And my name is Mo. [00:37]
-
-Carl: Yeah, more on that later. [00:51]
-
-## New releases
-
-Carl: Let's jump into some new releases. [00:52]
diff --git a/tests/fixtures/legacy-2026-05.md b/tests/fixtures/legacy-2026-05.md
deleted file mode 100644
index 96d66d9..0000000
--- a/tests/fixtures/legacy-2026-05.md
+++ /dev/null
@@ -1,39 +0,0 @@
----
-title: "TMiR 2026-05: Who even is on the Core team anymore, TanStack got pwn'd bad"
-date: 2026-05-28
-time: 2pm PT / 9pm GMT
-location: Main Stage on Reactiflux
-description: "Join Carl Vitullo and Mark Erikson as we discuss what the heck is going on at the React Core team."
-people: "[Carl](https://bsky.app/profile/vcarl.com), [Mark](https://bsky.app/profile/acemarke.dev)"
----
-
-<style>
-iframe {
-  position: sticky;
-  top: 0;
-}
-h2 {
-  scroll-margin-top: 188px;
-}
-</style>
-
-<iframe width="100%" height="180" frameborder="no" scrolling="no" seamless="" src="https://share.transistor.fm/e/dd8e79de"></iframe>
-
-- [[00:00:55](#some-podcast-meta)] Some podcast meta
-- [[00:01:49](#new-releases)] New Releases
-  - [[00:01:51](#ts-v7-beta)] [TS v7 beta](https://devblogs.microsoft.com/typescript/announcing-typescript-7-0-beta/)
-- [[01:08:39](#outro)] Outro
-
-# Interview
-
-**Carl Vitullo:** Thank you for joining us. We're coming to you live from Reactiflux. [00:00:00]
-
-## Some podcast meta
-
-**Carl Vitullo:** Okay, let's get into it. [00:00:55]
-
-Stuff has changed. Stuff will continue to change. Just like life. [00:01:40]
-
-## New Releases
-
-**Carl Vitullo:** But yeah, okay, into some new releases. [00:01:49]
diff --git a/tests/headings.test.ts b/tests/headings.test.ts
index adc4a82..a4cb258 100644
--- a/tests/headings.test.ts
+++ b/tests/headings.test.ts
@@ -45,9 +45,24 @@ test("items with no later paragraph, or sharing a target, are skipped", () => {
 
 test("paragraphs without a timestamp are never a target", () => {
   assert.deepEqual(
-    insertHeadings(["no time here", "later [00:02:00]"], [
-      { time: "00:00:00", title: "Intro" },
-    ]),
+    insertHeadings(
+      ["no time here", "later [00:02:00]"],
+      [{ time: "00:00:00", title: "Intro" }],
+    ),
     ["no time here", "## Intro", "later [00:02:00]"],
   );
 });
+
+test("an out-of-order outline item doesn't overwrite a heading already placed", () => {
+  assert.deepEqual(
+    insertHeadings(
+      ["p0 [00:00:10]", "p1 [00:01:00]", "p2 [00:05:00]"],
+      [
+        { time: "00:04:00", title: "Late" },
+        { time: "00:00:05", title: "Early" },
+        { time: "00:04:30", title: "Late2" },
+      ],
+    ),
+    ["## Early", "p0 [00:00:10]", "p1 [00:01:00]", "## Late", "p2 [00:05:00]"],
+  );
+});
diff --git a/tests/ingest.test.ts b/tests/ingest.test.ts
index 46a6b5c..2f74db4 100644
--- a/tests/ingest.test.ts
+++ b/tests/ingest.test.ts
@@ -4,23 +4,40 @@ import { readFileSync } from "node:fs";
 import { parseFeed, slugFromTitle, applyFeedItem } from "../scripts/ingest.ts";
 import { parseEpisode } from "../src/content/parse.ts";
 
-const xml = readFileSync(new URL("./fixtures/feed.xml", import.meta.url), "utf8");
+const xml = readFileSync(
+  new URL("./fixtures/feed.xml", import.meta.url),
+  "utf8",
+);
 
 test("slugFromTitle handles every TMiR title style and rejects the rest", () => {
-  assert.equal(slugFromTitle("TMiR 2026-05: Who even is on the Core team anymore"), "2026-05");
-  assert.equal(slugFromTitle("TMiR 2023-10: React Forget, Canary Releases"), "2023-10");
-  assert.equal(slugFromTitle("This Month in React – September 2023"), "2023-09");
+  assert.equal(
+    slugFromTitle("TMiR 2026-05: Who even is on the Core team anymore"),
+    "2026-05",
+  );
+  assert.equal(
+    slugFromTitle("TMiR 2023-10: React Forget, Canary Releases"),
+    "2023-10",
+  );
+  assert.equal(
+    slugFromTitle("This Month in React – September 2023"),
+    "2023-09",
+  );
   assert.equal(slugFromTitle("This Month In React – March 2023"), "2023-03");
   assert.equal(slugFromTitle("This Month in React (April 2023)"), "2023-04");
   assert.equal(
-    slugFromTitle("TMiR 2024-09 – Async Components??, a React 19 cheatsheet, static Hermes, and trademarks drama"),
+    slugFromTitle(
+      "TMiR 2024-09 – Async Components??, a React 19 cheatsheet, static Hermes, and trademarks drama",
+    ),
     "2024-09",
   );
   assert.equal(
     slugFromTitle("Mark & Carl talk with Swizec Teller about using AI at work"),
     "2026-04",
   );
-  assert.equal(slugFromTitle("Office Hours – States of Burnout with Jenny Truong"), null);
+  assert.equal(
+    slugFromTitle("Office Hours – States of Burnout with Jenny Truong"),
+    null,
+  );
   assert.equal(slugFromTitle("Behind the React Documentary"), null);
 });
 
@@ -48,7 +65,8 @@ test("parseFeed extracts TMiR items only", () => {
         img: "https://img.transistorcdn.com/carl.jpg",
       },
     ],
-    bskyPostUrl: "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
+    bskyPostUrl:
+      "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
   });
   assert.equal(items[1].slug, "2023-09");
   assert.equal(items[1].people.length, 1);
@@ -95,3 +113,41 @@ test("applyFeedItem writes only ingest-owned fields and is idempotent", () => {
   // atUri belongs to publish-atproto; ingest must leave it exactly as it found it
   assert.equal(ep.atUri, "at://did:plc:example/site.standard.document/2026-05");
 });
+
+test("parseFeed accepts HH:MM:SS and MM:SS itunes:duration", () => {
+  const feed = (duration: string) =>
+    [
+      "<item>",
+      "<title>TMiR 2026-06: Colon durations</title>",
+      '<enclosure url="https://media.transistor.fm/abc12345/hash.mp3" type="audio/mpeg"/>',
+      `<itunes:duration>${duration}</itunes:duration>`,
+      "</item>",
+    ].join("\n");
+
+  assert.equal(parseFeed(feed("1:09:58"))[0]?.duration, 4198);
+  assert.equal(parseFeed(feed("09:58"))[0]?.duration, 598);
+  assert.equal(parseFeed(feed("4198"))[0]?.duration, 4198);
+});
+
+test("applyFeedItem does not clobber curated people with an empty feed list", () => {
+  const file = [
+    "---",
+    'title: "TMiR 2026-05: test"',
+    "date: 2026-05-28",
+    'description: "d"',
+    "people:",
+    "  - name: Carl Vitullo",
+    "    role: Producer",
+    "---",
+    "",
+    "# Transcript",
+    "",
+    "**Carl Vitullo:** Hello. [00:00:55]",
+    "",
+  ].join("\n");
+
+  const item = { ...parseFeed(xml)[0], people: [] };
+  const ep = parseEpisode(applyFeedItem(file, item), "2026-05");
+  assert.equal(ep.people.length, 1);
+  assert.equal(ep.people[0].name, "Carl Vitullo");
+});
diff --git a/tests/jsonld.test.ts b/tests/jsonld.test.ts
new file mode 100644
index 0000000..ac3ac66
--- /dev/null
+++ b/tests/jsonld.test.ts
@@ -0,0 +1,14 @@
+import { test } from "node:test";
+import assert from "node:assert/strict";
+import { jsonLd } from "../src/content/jsonld.ts";
+
+test("a </script> in the data cannot close the block", () => {
+  const out = jsonLd({
+    description: "beware </script><script>alert(1)</script>",
+  });
+  assert.ok(!out.includes("<"));
+  assert.equal(
+    JSON.parse(out).description,
+    "beware </script><script>alert(1)</script>",
+  );
+});
diff --git a/tests/load.test.ts b/tests/load.test.ts
index 3a7657f..91aeb1b 100644
--- a/tests/load.test.ts
+++ b/tests/load.test.ts
@@ -13,18 +13,12 @@ test("loadEpisodes reads and parses the content directory", async () => {
   assert.equal(episodes[0].sections.length, 2);
 });
 
-test("loadEpisodes sorts by date descending", async () => {
-  const dates = (await loadEpisodes()).map((e) => e.date);
-  assert.deepEqual(dates, [...dates].sort().reverse());
-});
-
 test("getEpisode finds by slug and misses cleanly", async () => {
   assert.equal((await getEpisode("2026-05"))?.slug, "2026-05");
   assert.equal(await getEpisode("1999-01"), undefined);
 });
 
 test("loadEpisodes breaks a date tie by slug descending", async () => {
-  const { loadEpisodes: load } = await import("../src/content/load.ts");
   const tied = [
     { slug: "2026-06", date: "2026-06-24" },
     { slug: "2026-07", date: "2026-06-24" },
@@ -34,5 +28,4 @@ test("loadEpisodes breaks a date tie by slug descending", async () => {
     tied.sort(episodeOrder).map((e) => e.slug),
     ["2026-07", "2026-06", "2026-05"],
   );
-  assert.ok(await load());
 });
diff --git a/tests/migrate.test.ts b/tests/migrate.test.ts
deleted file mode 100644
index 62fff8e..0000000
--- a/tests/migrate.test.ts
+++ /dev/null
@@ -1,307 +0,0 @@
-import test from "node:test";
-import assert from "node:assert/strict";
-import { readFileSync } from "node:fs";
-import {
-  convert,
-  mergeExistingFrontMatter,
-  normalizeSpeakerName,
-  slugFromFilename,
-} from "../scripts/migrate.ts";
-import { parseEpisode, splitFile } from "../src/content/parse.ts";
-
-const fixture = (name: string) =>
-  readFileSync(new URL(`./fixtures/${name}`, import.meta.url), "utf8");
-
-test("slugFromFilename maps both filename styles", () => {
-  assert.equal(slugFromFilename("tmir-2024-03.md"), "2024-03");
-  assert.equal(slugFromFilename("tmir-march-2023.md"), "2023-03");
-  assert.equal(slugFromFilename("tmir-april-2023.md"), "2023-04");
-  assert.equal(slugFromFilename("tmir-dec-2023.md"), "2023-12");
-  assert.equal(slugFromFilename("tmir-nov-2023.md"), "2023-11");
-  assert.equal(slugFromFilename("tmir-october-2023.md"), "2023-10");
-  assert.equal(slugFromFilename("ali-spittel.md"), null);
-});
-
-test("2023 format: iframe stripped, outline synthesized from headings", () => {
-  const { text, warnings } = convert(fixture("legacy-2023-march.md"));
-  assert.ok(!text.includes("<iframe"));
-  assert.ok(
-    text.includes(
-      "- [[00:02:10](#retire-spa-and-mpa)] Retire SPA and MPA?\n" +
-        "- [[00:03:07](#create-react-app-no-longer-officially-recommended)] Create React App no longer officially recommended\n",
-    ),
-    text,
-  );
-  assert.ok(text.includes("\n# Transcript\n"));
-  assert.ok(
-    text.includes("**Carl Vitullo:** Thanks everyone for joining us for this month in React. [00:00:16]"),
-  );
-  assert.deepEqual(warnings, []);
-
-  const ep = parseEpisode(text, "2023-03");
-  assert.equal(ep.sections.length, 3); // Intro + 2
-  assert.equal(ep.sections[0].title, "Intro");
-  assert.equal(ep.outline.length, 2);
-  assert.equal(ep.outline[0].anchor, ep.sections[1].anchor);
-});
-
-test("2024 format: link list dropped, leading timestamps moved to the end", () => {
-  const { text } = convert(fixture("legacy-2024-03.md"));
-  assert.ok(!text.includes("[Spotify]("));
-  assert.ok(!text.includes("<iframe"));
-  assert.ok(
-    text.includes("- Quick Hits\n  - [[00:00:39](#layoffsfyi)] [Layoffs.fyi](https://layoffs.fyi/)\n"),
-    text,
-  );
-  assert.ok(
-    text.includes("**Carl Vitullo:** I'm going to start off with some layoffs news. [00:00:39]"),
-  );
-  assert.ok(text.includes("This month has about 7, 200 laid off so far. [00:00:58]"));
-
-  const ep = parseEpisode(text, "2024-03");
-  assert.equal(ep.sections.length, 3); // Intro, Layoffs, New releases
-  assert.equal(ep.sections[0].title, "Intro");
-  assert.equal(ep.sections[0].segments[0].speaker, "Carl Vitullo");
-  assert.equal(ep.sections[1].title, "Layoffs (seems better!)");
-  assert.equal(ep.sections[1].time, "00:00:39");
-  assert.equal(ep.sections.at(-1)?.title, "New releases");
-});
-
-test("2025-01 format: style block dropped, outline anchors re-derived", () => {
-  const { text } = convert(fixture("legacy-2025-01.md"));
-  assert.ok(!text.includes("<style>"));
-  assert.ok(!text.includes("<iframe"));
-  assert.ok(
-    text.includes(
-      "- [[00:01:29](#job-market-fred-data-layoffsfyi)] Job market: [FRED data](https://fred.stlouisfed.org/series/IHLIDXUSTPSOFTDEVE), [Layoffs.fyi](https://layoffs.fyi/)\n",
-    ),
-    text,
-  );
-  const ep = parseEpisode(text, "2025-01");
-  assert.equal(ep.outline[0].anchor, "job-market-fred-data-layoffsfyi");
-  assert.equal(ep.sections[1].anchor, "job-market-fred-data-layoffsfyi");
-});
-
-test("2025-06 format: heading timestamps moved to the outline, missing times warned", () => {
-  const { text, warnings } = convert(fixture("legacy-2025-06.md"));
-  assert.ok(text.includes("## job market\n"), text);
-  assert.ok(!text.includes("## [00:52]"));
-  assert.ok(text.includes("- Job market [FRED data]("));
-  assert.ok(
-    warnings.some((w) => w.includes("outline item without timestamp")),
-    JSON.stringify(warnings),
-  );
-  assert.ok(
-    warnings.some((w) => w.includes("dropped prose before outline")),
-    JSON.stringify(warnings),
-  );
-
-  const ep = parseEpisode(text, "2025-06");
-  assert.equal(ep.sections[1].title, "job market");
-  assert.equal(ep.sections[1].time, "00:00:52");
-  assert.equal(ep.sections[1].segments[0].speaker, "Carl Vitullo");
-  assert.equal(ep.outline[0].time, undefined);
-});
-
-test("2026-05 format: already canonical apart from the embed and marker", () => {
-  const { text, warnings } = convert(fixture("legacy-2026-05.md"));
-  assert.ok(!text.includes("<style>"));
-  assert.ok(!text.includes("<iframe"));
-  assert.ok(!text.includes("# Interview"));
-  assert.ok(text.includes("\n# Transcript\n"));
-  assert.deepEqual(warnings, []);
-
-  const ep = parseEpisode(text, "2026-05");
-  assert.equal(ep.outline.length, 3);
-  assert.equal(ep.outline[1].children.length, 1);
-  assert.equal(ep.sections.length, 3);
-  assert.equal(ep.sections[0].title, "Intro");
-});
-
-test("2025-09 format: unbolded speaker labels are bolded and normalized to full names", () => {
-  const { text, warnings } = convert(fixture("legacy-2025-09.md"));
-  assert.ok(!text.includes("<style>"));
-  assert.ok(text.includes("**Carl Vitullo:** Alright. Hello everyone."));
-  assert.ok(text.includes("**Mark Erikson:** Hi, I'm Mark."));
-  assert.ok(text.includes("**Mo Javad:** And my name is Mo."));
-  assert.deepEqual(warnings, []);
-
-  const ep = parseEpisode(text, "2025-09");
-  assert.equal(ep.sections[0].title, "Intro");
-  assert.deepEqual(
-    ep.sections[0].segments.map((s) => s.speaker),
-    ["Carl Vitullo", "Mark Erikson", "Mo Javad", "Carl Vitullo"],
-  );
-  assert.equal(ep.sections[1].title, "New releases");
-});
-
-test("unbolded speaker recognition doesn't misfire on ordinary `Word:` prose", () => {
-  const raw =
-    "---\ntitle: t\ndate: 2025-01-01\ndescription: d\n---\n\n" +
-    "Note: something important happened. [00:00]\n\n" +
-    "**Carl Vitullo:** Real speaker line. [00:05]\n";
-  const { text, warnings } = convert(raw);
-  assert.ok(!text.includes("**Note:**"));
-  assert.ok(
-    warnings.some((w) => w.includes("Note: something important happened")),
-  );
-  assert.ok(text.includes("**Carl Vitullo:** Real speaker line."));
-});
-
-test("2023-10 format: headings synthesized from top-level outline items when the transcript has none", () => {
-  const { text, warnings } = convert(fixture("legacy-2023-10.md"));
-  assert.deepEqual(warnings, []);
-  assert.ok(text.includes("\n## Introduction\n"), text);
-  assert.ok(text.includes("\n## Second topic\n"), text);
-  // Skipped: lands on the same paragraph as "Introduction".
-  assert.ok(!text.includes("## Also zero"));
-  // Skipped: no paragraph reaches this time.
-  assert.ok(!text.includes("## Never happens"));
-  // The outline itself is untouched (all four items still present).
-  assert.equal((text.match(/^- \[\[/gm) ?? []).length, 4);
-
-  const ep = parseEpisode(text, "2023-10");
-  assert.equal(ep.sections.length, 2);
-  assert.equal(ep.sections[0].title, "Introduction");
-  assert.equal(ep.sections[0].segments.length, 2);
-  assert.equal(ep.sections[1].title, "Second topic");
-  assert.equal(ep.sections[1].segments.length, 2);
-});
-
-test("raw Descript ids are normalized to full names", () => {
-  const raw =
-    '---\ntitle: t\ndate: 2025-11-26\ndescription: d\n---\n\n2-vcarl: Hello! [00:00]\n\n1-acemarke: Hi, Mark here. [00:10]\n';
-  const { text, warnings } = convert(raw);
-  assert.ok(text.includes("**Carl Vitullo:** Hello!"));
-  assert.ok(text.includes("**Mark Erikson:** Hi, Mark here."));
-  assert.deepEqual(warnings, []);
-});
-
-test("converted output re-serializes to itself (idempotent)", () => {
-  for (const name of [
-    "legacy-2023-march.md",
-    "legacy-2024-03.md",
-    "legacy-2025-01.md",
-    "legacy-2025-06.md",
-    "legacy-2025-09.md",
-    "legacy-2023-10.md",
-    "legacy-2026-05.md",
-  ]) {
-    const once = convert(fixture(name)).text;
-    assert.equal(convert(once).text, once, name);
-  }
-});
-
-test("mergeExistingFrontMatter keeps ingest-owned fields and atUri", () => {
-  const regenerated = [
-    "---",
-    'title: "TMiR 2026-05: regenerated"',
-    "date: 2026-05-28",
-    'description: "fresh"',
-    "---",
-    "",
-    "- Outline",
-    "",
-    "# Transcript",
-    "",
-    "**Carl Vitullo:** hi [00:00:01]",
-    "",
-  ].join("\n");
-  const existing = [
-    "---",
-    'title: "TMiR 2026-05: stale"',
-    "date: 2026-05-28",
-    'description: "stale"',
-    "people:",
-    "  - name: Mark Erikson",
-    "    role: Host",
-    "transistorId: dd8e79de",
-    "audioUrl: https://media.transistor.fm/dd8e79de/x.mp3",
-    "duration: 4198",
-    "season: 3",
-    "episode: 5",
-    "bskyPostUrl: https://bsky.app/profile/did:plc:x/post/abc",
-    "atUri: at://did:plc:x/site.standard.document/2026-05",
-    "---",
-    "",
-    "old body",
-    "",
-  ].join("\n");
-
-  const merged = mergeExistingFrontMatter(regenerated, existing);
-  const { frontMatter, body } = splitFile(merged);
-  // Hand-written fields come from the regenerated file...
-  assert.equal(frontMatter.title, "TMiR 2026-05: regenerated");
-  assert.equal(frontMatter.description, "fresh");
-  assert.ok(body.includes("**Carl Vitullo:** hi [00:00:01]"));
-  // ...ingest-owned fields and atUri from the existing one.
-  assert.equal(frontMatter.audioUrl, "https://media.transistor.fm/dd8e79de/x.mp3");
-  assert.equal(frontMatter.transistorId, "dd8e79de");
-  assert.equal(frontMatter.duration, 4198);
-  assert.equal(frontMatter.season, 3);
-  assert.equal(frontMatter.episode, 5);
-  assert.equal(frontMatter.bskyPostUrl, "https://bsky.app/profile/did:plc:x/post/abc");
-  assert.equal(frontMatter.atUri, "at://did:plc:x/site.standard.document/2026-05");
-  assert.deepEqual(frontMatter.people, [{ name: "Mark Erikson", role: "Host" }]);
-});
-
-test("normalizeSpeakerName canonicalizes bolded labels, guests keep theirs", () => {
-  assert.equal(normalizeSpeakerName("Carl Vitullo (editing)"), "Carl Vitullo");
-  assert.equal(normalizeSpeakerName("Carl (editing)"), "Carl Vitullo");
-  assert.equal(normalizeSpeakerName("Carl"), "Carl Vitullo");
-  assert.equal(normalizeSpeakerName("1-vcarl"), "Carl Vitullo");
-  assert.equal(normalizeSpeakerName("Mark"), "Mark Erikson");
-  assert.equal(normalizeSpeakerName("2-acemarke"), "Mark Erikson");
-  assert.equal(normalizeSpeakerName("Mo"), "Mo Javad");
-  assert.equal(normalizeSpeakerName("Mo", new Set(["mo khazali", "mo"])), "Mo");
-  assert.equal(normalizeSpeakerName("Mo Khazali"), "Mo Khazali");
-  assert.equal(normalizeSpeakerName("Sebastien Lorber"), "Sebastien Lorber");
-});
-
-test("convert normalizes bolded speaker labels in the transcript", () => {
-  const raw = [
-    "---",
-    'title: "t"',
-    "date: 2025-01-01",
-    'description: "d"',
-    "---",
-    "",
-    "- [00:00] Intro",
-    "",
-    "**Carl:** first [00:00:01]",
-    "",
-    "**Carl Vitullo (editing):** aside [00:00:05]",
-    "",
-    "**Mark:** second [00:00:09]",
-    "",
-    "**Sebastien Lorber:** guest [00:00:20]",
-    "",
-  ].join("\n");
-  const { text } = convert(raw);
-  assert.ok(text.includes("**Carl Vitullo:** first"), text);
-  assert.ok(text.includes("**Carl Vitullo:** aside"), text);
-  assert.ok(text.includes("**Mark Erikson:** second"), text);
-  assert.ok(text.includes("**Sebastien Lorber:** guest"), text);
-});
-
-test("convert leaves bare Mo alone when Mo Khazali is the guest", () => {
-  const file = (guest: string) =>
-    [
-      "---",
-      'title: "t"',
-      "date: 2025-01-01",
-      'description: "d"',
-      "---",
-      "",
-      "- [00:00] Intro",
-      "",
-      "**Carl:** hello [00:00:01]",
-      "",
-      `**${guest}:** hi [00:00:09]`,
-      "",
-      "**Mo:** and me [00:00:20]",
-      "",
-    ].join("\n");
-  assert.ok(convert(file("Mo Khazali")).text.includes("**Mo:** and me"));
-  assert.ok(convert(file("Mark")).text.includes("**Mo Javad:** and me"));
-});
diff --git a/tests/og.test.ts b/tests/og.test.ts
new file mode 100644
index 0000000..989eb14
--- /dev/null
+++ b/tests/og.test.ts
@@ -0,0 +1,32 @@
+import { test } from "node:test";
+import assert from "node:assert/strict";
+import { cardTitle, monthYear, runtime } from "../scripts/og.ts";
+
+test("strips the boilerplate from every title shape in the archive", () => {
+  assert.equal(
+    cardTitle("TMiR 2026-07: React-alikes, GOVERNANCE, and state management"),
+    "React-alikes, GOVERNANCE, and state management",
+  );
+  assert.equal(
+    cardTitle("This Month in React, May 2024: Updates from React Conf"),
+    "Updates from React Conf",
+  );
+  assert.equal(
+    cardTitle("This Month in React (March 2023)"),
+    "This Month in React",
+  );
+  // Nothing to strip: an off-format title survives intact.
+  assert.equal(
+    cardTitle("Mark & Carl talk with Swizec Teller about using AI at work"),
+    "Mark & Carl talk with Swizec Teller about using AI at work",
+  );
+});
+
+test("formats runtime and publication month", () => {
+  assert.equal(runtime(3252), "54m");
+  assert.equal(runtime(4200), "1h 10m");
+  assert.equal(runtime(undefined), "");
+  // Late-in-month dates must not roll back a month through a local timezone.
+  assert.equal(monthYear("2023-03-01"), "March 2023");
+  assert.equal(monthYear("2023-03-31"), "March 2023");
+});
diff --git a/tests/parse.test.ts b/tests/parse.test.ts
index 2b7936c..1c5cd37 100644
--- a/tests/parse.test.ts
+++ b/tests/parse.test.ts
@@ -75,7 +75,10 @@ test("speaker carries forward across consecutive paragraphs", () => {
   assert.equal(segs[0].speaker, "Carl Vitullo");
   assert.equal(segs[0].time, "00:00:55");
   assert.equal(segs[1].speaker, "Carl Vitullo");
-  assert.equal(segs[1].text, "Stuff has changed. Stuff will continue to change. Just like life.");
+  assert.equal(
+    segs[1].text,
+    "Stuff has changed. Stuff will continue to change. Just like life.",
+  );
   assert.equal(segs[2].speaker, "Mark Erikson");
 });
 
@@ -198,7 +201,11 @@ test("serialize writes a changed field without disturbing the body", () => {
   frontMatter.duration = 5000;
   const out = serializeEpisodeFile(frontMatter, body);
   assert.match(out, /^duration: 5000$/m);
-  assert.ok(out.includes("**Mark Erikson:** Yeah, that sounds all too real. [00:01:44]"));
+  assert.ok(
+    out.includes(
+      "**Mark Erikson:** Yeah, that sounds all too real. [00:01:44]",
+    ),
+  );
 });
 
 const withFrontMatter = (body: string) =>
@@ -317,7 +324,10 @@ test("toSrt tolerates sections with no segments", () => {
     ),
     "x",
   );
-  assert.match(toSrt(parsed), /^1\n00:00:01,000 --> 00:00:06,000\nCarl Vitullo: hi\n$/);
+  assert.match(
+    toSrt(parsed),
+    /^1\n00:00:01,000 --> 00:00:06,000\nCarl Vitullo: hi\n$/,
+  );
 });
 
 test("markdown escapes and code spans are undone in titles and segments", () => {
@@ -343,3 +353,23 @@ test("markdown escapes and code spans are undone in titles and segments", () =>
     "the react-concurrent-store ponyfill, snake_case",
   );
 });
+
+test("a body that begins with the transcript marker still parses its sections", () => {
+  const parsed = parseEpisode(
+    withFrontMatter(
+      ["# Transcript", "", "**Carl Vitullo:** hello [00:00:01]", ""].join("\n"),
+    ),
+    "x",
+  );
+  assert.deepEqual(parsed.outline, []);
+  assert.equal(parsed.sections.length, 1);
+  assert.equal(parsed.sections[0].segments[0].text, "hello");
+});
+
+test("a missing or unparseable date fails at parse time, naming the episode", () => {
+  const noDate = "---\ntitle: t\ndescription: d\n---\n\n# Transcript\n";
+  assert.throws(() => parseEpisode(noDate, "2026-05"), /2026-05.*date/);
+  const badDate =
+    "---\ntitle: t\ndate: nonsense\ndescription: d\n---\n\n# Transcript\n";
+  assert.throws(() => parseEpisode(badDate, "2026-05"), /2026-05.*date/);
+});
diff --git a/tests/publish-atproto.test.ts b/tests/publish-atproto.test.ts
index e11a165..39b8b3b 100644
--- a/tests/publish-atproto.test.ts
+++ b/tests/publish-atproto.test.ts
@@ -29,14 +29,21 @@ test("outlineToText flattens the outline to indented 'Title — url' lines", ()
 
 test("bskyUrlToParts splits a post URL and rejects anything else", () => {
   assert.deepEqual(
-    bskyUrlToParts("https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x"),
+    bskyUrlToParts(
+      "https://bsky.app/profile/thismonthinreact.com/post/3lqz7abcd2k2x",
+    ),
     { actor: "thismonthinreact.com", rkey: "3lqz7abcd2k2x" },
   );
   assert.deepEqual(
-    bskyUrlToParts("https://bsky.app/profile/did:plc:abc123/post/3lqz7abcd2k2x"),
+    bskyUrlToParts(
+      "https://bsky.app/profile/did:plc:abc123/post/3lqz7abcd2k2x",
+    ),
     { actor: "did:plc:abc123", rkey: "3lqz7abcd2k2x" },
   );
-  assert.equal(bskyUrlToParts("https://bsky.app/profile/thismonthinreact.com"), null);
+  assert.equal(
+    bskyUrlToParts("https://bsky.app/profile/thismonthinreact.com"),
+    null,
+  );
   assert.equal(bskyUrlToParts("https://example.com/whatever"), null);
 });
 
diff --git a/tests/publish-transcript.test.ts b/tests/publish-transcript.test.ts
index 75da749..6791916 100644
--- a/tests/publish-transcript.test.ts
+++ b/tests/publish-transcript.test.ts
@@ -5,6 +5,7 @@ import {
   descriptToCanonical,
   replaceTranscript,
   addOutlineHeadings,
+  fetchDescriptTranscript,
 } from "../scripts/publish-transcript.ts";
 import { parseEpisode } from "../src/content/parse.ts";
 
@@ -62,7 +63,9 @@ test("replaceTranscript keeps front matter and outline, replaces the body below
   );
 
   assert.ok(!out.includes("stale text"));
-  assert.ok(out.includes("- [[00:00:55](#some-podcast-meta)] Some podcast meta"));
+  assert.ok(
+    out.includes("- [[00:00:55](#some-podcast-meta)] Some podcast meta"),
+  );
   assert.ok(out.includes("transistorId: dd8e79de"));
 
   const ep = parseEpisode(out, "2026-05");
@@ -122,3 +125,29 @@ test("addOutlineHeadings gives a Descript body sections from the file's outline"
     ["intro", "main-content"],
   );
 });
+
+test("fetchDescriptTranscript refuses a degenerate 200 body", async () => {
+  const real = globalThis.fetch;
+  const stub = (body: string) => {
+    globalThis.fetch = (async () =>
+      new Response(body, { status: 200 })) as typeof fetch;
+  };
+  try {
+    stub("");
+    await assert.rejects(
+      fetchDescriptTranscript("p", "t"),
+      /refusing to overwrite/,
+    );
+    stub("   \n\n  ");
+    await assert.rejects(
+      fetchDescriptTranscript("p", "t"),
+      /refusing to overwrite/,
+    );
+
+    const real200 = `[00:00] **1-vcarl:** ${"Hello everyone. ".repeat(60)}`;
+    stub(real200);
+    assert.equal(await fetchDescriptTranscript("p", "t"), real200);
+  } finally {
+    globalThis.fetch = real;
+  }
+});
diff --git a/tests/slug.test.ts b/tests/slug.test.ts
index 8068ba1..d4280fd 100644
--- a/tests/slug.test.ts
+++ b/tests/slug.test.ts
@@ -78,10 +78,15 @@ test("splitTitleLink separates a heading's plain text from its first URL", () =>
 
 test("unescapeMarkdown drops backslash escapes and unwraps code spans", () => {
   assert.equal(unescapeMarkdown("Redux \\\\\\+ signals"), "Redux + signals");
-  assert.equal(unescapeMarkdown("React Native \\\\\\<\\\\\\> Imgui"), "React Native <> Imgui");
+  assert.equal(
+    unescapeMarkdown("React Native \\\\\\<\\\\\\> Imgui"),
+    "React Native <> Imgui",
+  );
   assert.equal(unescapeMarkdown("snake\\_case"), "snake_case");
   assert.equal(
-    unescapeMarkdown("Third-party \\\\\\`react-concurrent-store\\\\\\` ponyfill"),
+    unescapeMarkdown(
+      "Third-party \\\\\\`react-concurrent-store\\\\\\` ponyfill",
+    ),
     "Third-party react-concurrent-store ponyfill",
   );
   assert.equal(unescapeMarkdown("plain `code` span"), "plain code span");
@@ -91,10 +96,15 @@ test("unescapeMarkdown drops backslash escapes and unwraps code spans", () => {
 
 test("flattenLinks and slug unescape too", () => {
   assert.equal(
-    flattenLinks("[TS 6.0 may enable \\\\\\`strict\\\\\\` by default](https://x.example)"),
+    flattenLinks(
+      "[TS 6.0 may enable \\\\\\`strict\\\\\\` by default](https://x.example)",
+    ),
     "TS 6.0 may enable strict by default",
   );
-  assert.equal(slug("Mark's React-Redux \\\\\\+ signals draft"), "marks-react-redux-signals-draft");
+  assert.equal(
+    slug("Mark's React-Redux \\\\\\+ signals draft"),
+    "marks-react-redux-signals-draft",
+  );
   assert.deepEqual(
     splitTitleLink("[React Native \\\\\\<\\\\\\> Imgui](https://x.example/1)"),
     { text: "React Native <> Imgui", url: "https://x.example/1" },
diff --git a/tests/srt.test.ts b/tests/srt.test.ts
index 891e4da..d319149 100644
--- a/tests/srt.test.ts
+++ b/tests/srt.test.ts
@@ -38,8 +38,41 @@ test("toSrt numbers cues, formats times, and prefixes the speaker", () => {
   );
 });
 
-test("the last cue runs five seconds", () => {
+test("cues never end before, or at, their own start", () => {
+  const raw = `---
+title: t
+date: 2026-05-28
+description: d
+---
+
+# Transcript
+
+**Carl Vitullo:** same second [00:01:00]
+
+**Mark Erikson:** also same second [00:01:00]
+
+**Carl Vitullo:** backwards [00:00:30]
+
+**Mark Erikson:** last [00:02:00]
+`;
   const srt = toSrt(parseEpisode(raw, "2026-05"));
-  assert.ok(srt.trimEnd().endsWith("Carl Vitullo: But yeah, okay, into some new releases."));
-  assert.ok(srt.includes("00:01:49,000 --> 00:01:54,000"));
+  const spans = [
+    ...srt.matchAll(/^(\d\d:\d\d:\d\d),000 --> (\d\d:\d\d:\d\d),000$/gm),
+  ];
+  assert.equal(spans.length, 4);
+  for (const [, start, end] of spans)
+    assert.ok(end > start, `${start} --> ${end}`);
+});
+
+test("every episode in content/ produces cues with a positive duration", async () => {
+  const { loadEpisodes } = await import("../src/content/load.ts");
+  const bad: string[] = [];
+  for (const episode of await loadEpisodes()) {
+    for (const [, start, end] of toSrt(episode).matchAll(
+      /^(\d\d:\d\d:\d\d),000 --> (\d\d:\d\d:\d\d),000$/gm,
+    )) {
+      if (end <= start) bad.push(`${episode.slug} ${start} --> ${end}`);
+    }
+  }
+  assert.deepEqual(bad, []);
 });
diff --git a/tests/time.test.ts b/tests/time.test.ts
index 0fe4f2e..28cea7a 100644
--- a/tests/time.test.ts
+++ b/tests/time.test.ts
@@ -23,3 +23,10 @@ test("isoDuration formats seconds as an ISO 8601 duration", () => {
   assert.equal(isoDuration(3600), "PT1H0S");
   assert.equal(isoDuration(0), "PT0S");
 });
+
+test("toSeconds rejects empty and over-long values", () => {
+  assert.equal(toSeconds(""), undefined);
+  assert.equal(toSeconds("   "), undefined);
+  assert.equal(toSeconds("00::30"), undefined);
+  assert.equal(toSeconds("1:2:3:4"), undefined);
+});
diff --git a/vite.config.ts b/vite.config.ts
index a912c38..44a2f8f 100644
--- a/vite.config.ts
+++ b/vite.config.ts
@@ -14,8 +14,12 @@ const slugs = readdirSync("content/episodes")
   .sort();
 
 const pages: { path: string; prerender: { enabled: boolean } }[] = [
-  ...slugs.map((slug) => ({ path: `/episodes/${slug}`, prerender: { enabled: true } })),
+  ...slugs.map((slug) => ({
+    path: `/episodes/${slug}`,
+    prerender: { enabled: true },
+  })),
   { path: "/feed.xml", prerender: { enabled: true } },
+  { path: "/sitemap.xml", prerender: { enabled: true } },
   ...slugs.map((slug) => ({
     path: `/episodes/${slug}/chapters.json`,
     prerender: { enabled: true },