[MeharPro]

Owner requests, 2026-11-12: usage limits on errors fixed live, then GenClass strictly observes (plans per project, #224, deployed); and "Yes, raise it" for the first-load budget.

What it does

• src/telemetry/transport.ts: the fetch transport hands back the collector's x-genclass-observe reply header ({ observe }, null when absent).
• src/telemetry/client.ts planReply: the header caps the runtime at observe and is remembered per app in localStorage["genclass:observe"] ("<token> <end of the UTC month ms>"), so the next page starts capped; a reply without it lifts the cap and clears the key. A reply the page cannot read (beacon, custom transport) changes nothing.
• src/runtime.ts capToObserve: one more cap in sessionMode (effectiveMode = min(mode, sample, breaker, plan, route)); one console warning. It only ever lowers the mode.

Rules

• Rule 1: observe changes nothing, so a correct app is never made worse.
• Rule 8: a dated exception (AGENTS.md, CONTRACT.md §17): the reply may lower the mode; nothing more is sent; with telemetry off there is no reply and no cap. TELEMETRY.md, both READMEs and the CHANGELOG say so.
• Rule 3: no model-visible text changes. Rule 5: the month's end comes from the injected Clock.

First load

Growth over main at e4f2eaa (Node 22.23, test/first-load-size.mjs): /auto 108.939 → 109.239 KB (0.300 KB), the main entry 100.672 → 100.964 KB (0.292 KB). test/bundle.test.ts limits raised by 0.3 KB each (109.3 / 101.0 KB), owner OK 2026-11-12; RESULTS.md §14 row added.

Tests

Runtime: typecheck clean, build, 1273 passed / 14 skipped (model parity) with --exclude test/browser/** --exclude test/review-perf.test.ts; two new tests in test/telemetry.test.ts. No review test edited.

Not in this PR: a runtime release (version bump and the v* tag) and a genclass CLI release that installs it; both wait for the owner.

🤖 Generated with Claude Code

[genclass-dev]

GenClass runtime review

2 findings in this pull request's tests at 47028aa · 2 reproducible · 5 tests read

#	Status	Finding	Test	Seen
1	New	Witness, overwrote-newer · GET /api/search	[chromium] search.spec.ts > search > buggy page: results of the last query	2 of 2 seeds, reproducible
2	New	Witness, overwrote-newer · GET /api/search	search.spec.ts > search > buggy page: results of the last query	5 of 5 seeds, reproducible

Reply @genclass-dev replay 1 for the steps to reproduce a finding, or @genclass-dev fix 1 for a brief a coding agent can work from.

The race report, test by test

Race report

2 signatures in 2 tests, 2 reproducible (flagged in 2 or more seeds of the same test) · 19 files read · mode observe

	signatures	reproducible
Duplicate submits	0	0
Stale responses	0	0
Failing endpoints	0	0
Other detections	0	0
Witnesses (facts measured without the model)	2	2

[chromium] search.spec.ts > search > buggy page: results of the last query

2 seeds (1, 2) · 1 signature · chaos: latency 0–600 ms, reorder

Witnesses (facts measured without the model)

• Reproducible: 2 of 2 seeds (1, 2) · GET /api/search · witness:overwrote-newer

  The response of GET /api/search#​1 wrote search.results over what newer GET /api/search#​2, started 0.08s after GET /api/search#​1, had written (2 items ["re one", "re two"] → 2 items ["r one", "r two"]).

search.spec.ts > search > buggy page: results of the last query

5 seeds (1, 2, 3, 4, 5) · 1 signature · chaos: latency 0–600 ms, reorder

Witnesses (facts measured without the model)

• Reproducible: 5 of 5 seeds (1, 2, 3, 4, 5) · GET /api/search · witness:overwrote-newer

  The response of GET /api/search#​1 wrote search.results over what newer GET /api/search#​4, started 0.25s after GET /api/search#​1, had written (2 items ["reac one", "reac two"] → 2 items ["r one", "r two"]).

3 tests with nothing flagged

• [chromium] search.spec.ts > search > fixed page: results of the last query (2 seeds: 1, 2)
• search.spec.ts > search > fixed page: results of the last query (5 seeds: 1, 2, 3, 4, 5)
• search.spec.ts > search > latency only > buggy page, latency only: results of the last query (5 seeds: 1, 2, 3, 4, 5)

---

Lines under "Witnesses (facts measured without the model)" are happens-before facts the runtime measured in the page without the model, one per signature (witness kind, operation): overwrote-newer, a response wrote a field over what a newer request or the user had written since the request started; identical-in-flight, two identical requests in flight together from separate user actions. They carry no confidence: a fact is listed when it happened. Reproducible: seen in 2 or more seeds of the same test. Whether a fact is a bug is for you to decide (some overwrites are intended; many are repaired a moment later).

---

_{genclass-dev[bot] read 19 race files from the artifacts genclass-races-unmodified, genclass-races of this pull request's workflow run, never your source code. What the lines mean · Setup and settings · reply @genclass-dev help}