fix(unic-pr-review): Work Item discovery always returns empty (fetcher never populates workItemRefs)

apps/claude-code/unic-pr-review/.claude-plugin/marketplace.json

@@ -21,7 +21,7 @@
       "name": "unic-pr-review",
       "source": "./",
       "tags": ["productivity", "code-review", "azure-devops"],
-      "version": "2.1.9"
+      "version": "2.1.10"
     }
   ]
 }

apps/claude-code/unic-pr-review/.claude-plugin/plugin.json

@@ -15,5 +15,5 @@
   "keywords": ["pr-review", "azure-devops", "jira", "confluence", "code-review", "unic"],
   "license": "LGPL-3.0-or-later",
   "name": "unic-pr-review",
-  "version": "2.1.9"
+  "version": "2.1.10"
 }

apps/claude-code/unic-pr-review/CHANGELOG.md

@@ -16,6 +16,17 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Fixed
 - (none)
 
+## [2.1.10] — 2026-06-16
+
+### Breaking
+- (none)
+
+### Added
+- (none)
+
+### Fixed
+- ADO Fetcher now fetches linked Work Items via `pullrequestworkitems` endpoint (Step 1.5) and populates `prMetadata.workItemRefs`; previously the fetcher never called this endpoint so `discoverWorkItems()` always returned `[]` and the Intent Check was silently skipped on every PR with linked Work Items (#247). Step 1.5 discriminates three fetch outcomes so a zero-exit-but-malformed response can no longer masquerade as "no Work Items linked"
+
 ## [2.1.9] — 2026-06-09
 
 ### Breaking

apps/claude-code/unic-pr-review/agents/ado-fetcher.md

@@ -38,6 +38,30 @@ If the command exits non-zero, emit this error and stop:
 { "error": "fetch-failed", "step": 1, "resource": "pullrequests", "message": "<stderr>" }
 ```
 
+### Step 1.5 — Fetch linked Work Items
+
+The `pullrequests` endpoint does not include Work Item links. Fetch them from the dedicated endpoint:
+
+```sh
+az devops invoke --area git --resource pullrequestworkitems \
+  --route-parameters organization="<orgUrl>" project="<project>" repositoryId="<repo>" pullRequestId="<prId>" \
+  --http-method GET --api-version 7.0 --output json
+```
+
+Store stdout as `WI_LIST` and capture the exit code. Distinguish three outcomes — never let a failure masquerade as the legitimate "no Work Items linked" case (the silent false negative this step exists to prevent):
+
+- **Exit zero and stdout parses to an object with a `value` array** → set `PR_METADATA.workItemRefs = WI_LIST.value`. An empty `value` legitimately means no Work Items are linked — no warning.
+- **Exit non-zero** → set `PR_METADATA.workItemRefs = []` and add the warning below (`<reason>` = `Error: <stderr>`).
+- **Exit zero but stdout is empty, unparseable, or has no `value` array** → treat as a fetch failure, not zero Work Items: set `PR_METADATA.workItemRefs = []` and add the warning below (`<reason>` = `unexpected response shape — do not assume zero Work Items`).
+
+In both failure branches add this warning (it belongs to the same `warnings` array emitted in Step 6):
+
+```
+"pullrequestworkitems fetch failed — Work Item discovery skipped, Intent Check will be omitted. <reason>"
+```
+
+Do not stop; continue to Step 2.
+
 ### Step 2 — Fetch Revisions (iterations)
 
 ```sh
@@ -232,7 +256,10 @@ Emit exactly one JSON object — no prose, no markdown, no footer (replace each
 
 ```json
 {
-  "prMetadata": "<PR_METADATA object>",
+  "prMetadata": {
+    "pullRequestId": 42,
+    "workItemRefs": [{ "id": "101", "url": "https://dev.azure.com/org/project/_apis/wit/workitems/101" }]
+  },
   "revisions": "<REVISIONS object>",
   "threads": "<THREADS object>",
   "changedFiles": ["path/to/file.ts"],
@@ -247,4 +274,6 @@ Emit exactly one JSON object — no prose, no markdown, no footer (replace each
 }
 ```
 
+`prMetadata` is the raw ADO `pullrequests` response enriched with `workItemRefs` from Step 1.5. `workItemRefs` is always an array (empty `[]` when no Work Items are linked or the fetch failed).
+
 `mode` is one of `"first-review"`, `"re-review"`, `"first-review-fallback"`. `priorRevisionId` and `priorIteration` are `null` except in `re-review` mode (where they carry `PRIOR_SIG.priorRevisionId` / `PRIOR_SIG.priorIteration`). `deltaRawDiff` is the delta diff string (empty in first-review modes). `priorFindings` is an array of `{ threadId, filePath, startLine, severity, title }` objects (empty except in `re-review` mode), where `threadId` is the number id of the ADO Thread carrying that prior finding's bot comment — it is what the Re-review Coordinator keys all thread mapping on. `diffUnavailable` is `false` when a real diff was computed (re-review always, first-review/first-review-fallback when inside a matching clone and the git diff succeeds) and `true` when a diff could not be obtained (no matching clone, missing commonRefCommit or sourceRefCommit, git diff failure, or an empty diff despite a non-empty changedFiles). `warnings` is an array of strings for any non-fatal issues. Never emit `hardStop` — the orchestrator handles all write decisions.

apps/claude-code/unic-pr-review/docs/adr/0001-multi-source-intent-with-shared-atlassian-credentials.md

@@ -24,9 +24,11 @@ Reuse `~/.unic-confluence.json` as the shared Atlassian Credential File for both
 ## Amendment (2026-06)
 
 Work-item discovery is a **Provider contract**. Each Source Platform Provider exposes
-`discoverWorkItems(prMetadata) → [{ id, type, url, raw }]`. For ADO, this reads the
-PR's native `workItemRefs` field (not regex-scraping the description); for future
-GitHub/GitLab Providers it will use their respective native linkages. The Intent Checker
-stays Source-Platform-agnostic: it consumes the normalised list regardless of origin.
-This separates "where did the Work Items come from?" (Provider contract) from "what
-intent do they express?" (Intent Checker responsibility).
+`discoverWorkItems(prMetadata) → [{ id, type, url, raw }]`. For ADO, the Fetcher
+populates `prMetadata.workItemRefs` from the dedicated `pullrequestworkitems` endpoint
+(Step 1.5 of the ADO Fetcher — the `pullrequests` response does not include Work Item
+links); `discoverWorkItems` then reads that field (never regex-scraping the description).
+For future GitHub/GitLab Providers it will use their respective native linkage endpoints.
+The Intent Checker stays Source-Platform-agnostic: it consumes the normalised list
+regardless of origin. This separates "where did the Work Items come from?" (Provider
+contract) from "what intent do they express?" (Intent Checker responsibility).

apps/claude-code/unic-pr-review/docs/adr/0016-human-threads-as-read-only-context.md

@@ -6,15 +6,15 @@
 
 The ADO Fetcher fetches every PR Thread but classifies them on a single axis: does `comments[0].content` carry the Iteration Marker (a Bot Thread, ADR-0006) or not. In `first-review` mode the fetched Threads are then dropped — no step reads them — and in `re-review` only Bot Threads feed the Re-review Coordinator's `priorFindings`. Human review discussion is invisible to a Review in every Mode.
 
-This surfaced in a head-to-head against the legacy `pr-review` plugin on ADO PR #5570. That plugin read the five Human Threads on the PR, cross-referenced a Finding to a still-open one, and noted a Thread marked *fixed* whose issue was in fact still present ("re-verify before merge"). `unic-pr-review` could surface none of this — a structural gap, not a fetch failure.
+This surfaced in a head-to-head against the legacy `pr-review` plugin on ADO PR #5570. That plugin read the five Human Threads on the PR, cross-referenced a Finding to a still-open one, and noted a Thread marked _fixed_ whose issue was in fact still present ("re-verify before merge"). `unic-pr-review` could surface none of this — a structural gap, not a fetch failure.
 
-The tempting fix — let a resolved Human Thread suppress or down-rank a matching Finding — is the mirror image of the work-item-discovery false negative (issue #247): where that bug made the Plugin *under-gather intent*, suppression would make it *under-report code issues*. A wrongly-resolved Thread would silently hide a real defect.
+The tempting fix — let a resolved Human Thread suppress or down-rank a matching Finding — is the mirror image of the work-item-discovery false negative (issue #247): where that bug made the Plugin _under-gather intent_, suppression would make it _under-report code issues_. A wrongly-resolved Thread would silently hide a real defect.
 
 ## Decision
 
 Human Threads are **read-only context**, surfaced after the aspect fan-out and never used to suppress a Finding.
 
-1. **Classification.** The Fetcher gains a human/system split alongside the existing bot detection: *Bot Thread* = Iteration Marker (ADR-0006); *System Thread* = `comments[0].commentType === "system"` (ref updates, votes, policy); *Human Thread* = neither. Only Human Threads are surfaced; System Threads are never shown to the Reviewer.
+1. **Classification.** The Fetcher gains a human/system split alongside the existing bot detection: _Bot Thread_ = Iteration Marker (ADR-0006); _System Thread_ = `comments[0].commentType === "system"` (ref updates, votes, policy); _Human Thread_ = neither. Only Human Threads are surfaced; System Threads are never shown to the Reviewer.
 
 2. **Context-only, never suppress.** A Finding is never dropped or down-ranked because a Human Thread exists. Confidence < 60 remains the sole filter (ADR-0002).
 

apps/claude-code/unic-pr-review/package.json

@@ -22,5 +22,5 @@
     "verify:changelog": "unic-verify-changelog"
   },
   "type": "module",
-  "version": "2.1.9"
+  "version": "2.1.10"
 }

apps/claude-code/unic-pr-review/providers/azure_devops/README.md

@@ -25,7 +25,9 @@ Throws `Not an ADO PR URL: <url>` when the URL does not match `prUrlPattern`.
 
 ## `discoverWorkItems(prMetadata)`
 
-Reads the PR's **native** `workItemRefs` field and normalises each entry to
+Reads the `workItemRefs` that the ADO Fetcher populates on `prMetadata` from the
+dedicated `pullrequestworkitems` endpoint (Step 1.5 — the `pullrequests` response
+does not carry Work Item links), and normalises each entry to
 `{ id, type: "ado-work-item", url, raw }`. It never regex-scrapes the PR
 description — work-item discovery is a Provider contract (ADR-0001 amendment).
 Returns `[]` when `workItemRefs` is empty or absent.
@@ -39,8 +41,9 @@ Returns `[]` when `workItemRefs` is empty or absent.
 
 ## Adding fixtures
 
-PR-metadata fixtures live in `fixtures/`. They mirror the shape returned by
-`az devops invoke --area git --resource pullrequests`. Add a new fixture file and
+PR-metadata fixtures live in `fixtures/`. They mirror the enriched `prMetadata`
+shape — the `az devops invoke --area git --resource pullrequests` response with
+`workItemRefs` grafted on by the Fetcher (Step 1.5). Add a new fixture file and
 reference it from `tests/provider.test.mjs` via the `fixture(name)` helper. The
 `fixtures/ado-cli-inventory.json` file catalogues every `az devops invoke` call
 the ADO Fetcher agent emits; the root `tests/ado-cli-smoke.test.mjs` asserts the

apps/claude-code/unic-pr-review/providers/azure_devops/fixtures/ado-cli-inventory.json

@@ -1,6 +1,12 @@
 {
   "invokeCommands": [
     { "area": "git", "resource": "pullrequests", "httpMethod": "GET", "description": "PR metadata" },
+    {
+      "area": "git",
+      "resource": "pullrequestworkitems",
+      "httpMethod": "GET",
+      "description": "Work Items linked to the PR (Step 1.5 — separate endpoint, not in pullrequests response)"
+    },
     {
       "area": "git",
       "resource": "pullrequestiterations",

apps/claude-code/unic-pr-review/providers/azure_devops/provider.mjs

@@ -41,10 +41,11 @@ export const agents = {
 }
 
 /**
- * Read the PR's native Work Item field (`workItemRefs`). Never regex-scrapes the
- * description — this is the ADR-0001 amendment contract.
+ * Read the Work Item refs that the ADO Fetcher populated on `prMetadata` from the
+ * PR's linked work-item endpoint (`pullrequestworkitems`, Step 1.5). Never
+ * regex-scrapes the description — this is the ADR-0001 amendment contract.
  *
- * @param {{ workItemRefs?: Array<{ id: string, url: string }> }} prMetadata
+ * @param {{ workItemRefs?: Array<{ id: string | number, url: string }> }} prMetadata
  * @returns {Array<{ id: string, type: string, url: string, raw: object }>}
  * @throws {Error} when `prMetadata` is not a PR-metadata object (guards against a
  *   malformed stdin payload silently yielding zero Work Items)

apps/claude-code/unic-pr-review/providers/azure_devops/tests/provider.test.mjs

@@ -83,4 +83,49 @@ describe('discoverWorkItems', () => {
 			assert.equal(item.type, 'ado-work-item')
 		}
 	})
+	it('normalises integer wire-format ids to strings', () => {
+		// The live pullrequestworkitems endpoint returns integer ids; fixtures use strings.
+		// discoverWorkItems must coerce via String(ref.id) so downstream id handling is stable
+		// regardless of wire shape — this exercises the integer branch the JSDoc `id: string | number`
+		// widening opened, which the fixture-shape tests below cannot reach.
+		const items = discoverWorkItems({
+			workItemRefs: [{ id: 42622, url: 'https://dev.azure.com/FZAG/_apis/wit/workitems/42622' }],
+		})
+		assert.equal(items.length, 1)
+		assert.equal(items[0].id, '42622')
+		assert.equal(typeof items[0].id, 'string')
+	})
+})
+
+// Fetcher contract: the ADO Fetcher (Step 1.5) must populate workItemRefs on prMetadata
+// from the pullrequestworkitems endpoint. These tests assert that the fixtures reflect
+// real fetcher output shape — workItemRefs is always present (never absent) because the
+// fetcher sets it explicitly (to [] when no WIs are linked or the fetch fails).
+describe('fetcher contract — prMetadata.workItemRefs is always fetcher-populated', () => {
+	it('pr-with-work-items fixture has workItemRefs with at least one entry', () => {
+		const meta = fixture('pr-with-work-items.json')
+		assert.ok(Array.isArray(meta.workItemRefs), 'workItemRefs must be an array (fetcher-populated)')
+		assert.ok(meta.workItemRefs.length > 0, 'fixture must have at least one work item ref')
+	})
+	it('pr-with-multiple-work-items fixture has workItemRefs with multiple entries', () => {
+		const meta = fixture('pr-with-multiple-work-items.json')
+		assert.ok(Array.isArray(meta.workItemRefs), 'workItemRefs must be an array (fetcher-populated)')
+		assert.ok(meta.workItemRefs.length > 1, 'fixture must have more than one work item ref')
+	})
+	it('pr-without-work-items fixture has workItemRefs as empty array (not absent)', () => {
+		const meta = fixture('pr-without-work-items.json')
+		assert.ok(Array.isArray(meta.workItemRefs), 'workItemRefs must be present even when empty')
+		assert.equal(meta.workItemRefs.length, 0)
+	})
+	it('each workItemRef has string id and url', () => {
+		// Note: fixtures use synthetic string ids ("101") for simplicity.
+		// The real ADO pullrequestworkitems endpoint returns integer ids (101).
+		// discoverWorkItems normalises both via String(ref.id), so runtime is safe;
+		// this assertion matches the fixture shape, not the wire format.
+		const meta = fixture('pr-with-work-items.json')
+		for (const ref of meta.workItemRefs) {
+			assert.equal(typeof ref.id, 'string', 'workItemRef.id must be a string')
+			assert.equal(typeof ref.url, 'string', 'workItemRef.url must be a string')
+		}
+	})
 })