Repository navigation
CI: a shorter critical path, a timeout on every job, one run per PR - #845
Merged
Merged
Conversation
Measured on main's recent runs (job p95 at most 6 minutes): - Twelve CI jobs waited on lint-and-typecheck, test or build without using anything they produced; e2e waited for build and then rebuilt everything itself, so it started 6 to 9 minutes into the run and finished last every time. Only the two pack jobs pass artifacts; they keep their needs, every other job starts at once. - No job had a timeout, so a hang (this repo's recurring bug class is processes that never exit) held a runner and kept a PR pending for up to 6 hours. Every CI job now stops at 20 minutes. - No PR workflow cancelled the run a new push replaced. ci, claims-alignment, codeql, lighthouse and publish-python now keep one run per pull request; main and tags are never cancelled. - Every PR wrote its own Docker layer cache, which only that PR can read; the repository's cache stood at 10.7 GB against a 10 GB limit. Only main writes it now. - Lighthouse (not a required check) runs when the dashboard, its server, the seed data or its budget change, and a failed seed fails it instead of scoring an empty dashboard. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub. |
Iris gate — 1 of 2 tripped
|
| Trace | Verdict | Basis | Rules, classes or missing inputs | Evidence |
|---|---|---|---|---|
d2dd6d74f5ccc21c05dfe2d4abeae042 |
failed | detector_veto + risk_over_loss |
no_pii, pii_leak, credential_leak | no_pii: AWS Access Key (output 45–65) |
| Verdict basis | Traces |
|---|---|
detector_veto |
2 |
clean |
1 |
Unjudged questions: task_completed (3), tool_use_correct (3) — a trace that did not carry what a rule needs.
tests/fixtures/ci-gate/traces.ndjson · 3 evaluated · dataset release-gate: 2 in the gate · exit 1 · what the bases mean
Iris gate — 1 stored, nothing tripped
|
| Verdict basis | Traces |
|---|---|
clean |
1 |
Unjudged questions: task_completed (1), tool_use_correct (1) — a trace that did not carry what a rule needs.
tests/fixtures/ci-gate/clean.ndjson · 1 evaluated · exit 0 · what the bases mean
irparent
added a commit
that referenced
this pull request
Oct 5, 2026
…its own #845 gave ci, claims-alignment, codeql, lighthouse and publish-python one concurrency group per pull request, cancelling a run its PR replaced. On main the group was the branch, so a merge queued behind a running one replaced the merge queued before it: two of main's commits on 10-05 never got their CI run. A pull request keeps its group; every other run (a push to main, a tag, a schedule) is grouped by its own run id, so none is replaced. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
irparent
added a commit
that referenced
this pull request
Oct 5, 2026
The 20-minute limit every job gained in #845 cancelled e2e on a pull request whose tests were passing: the browser download took 976 s against about a minute normally (Playwright's download server, not the change). The browsers are now cached per Playwright version, so most runs download nothing, and e2e, the one job that depends on someone else's server for a large download, gets 40 minutes; every other job keeps 20. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
irparent
added a commit
that referenced
this pull request
Oct 5, 2026
…its own #845 gave ci, claims-alignment, codeql, lighthouse and publish-python one concurrency group per pull request, cancelling a run its PR replaced. On main the group was the branch, so a merge queued behind a running one replaced the merge queued before it: two of main's commits on 10-05 never got their CI run. A pull request keeps its group; every other run (a push to main, a tag, a schedule) is grouped by its own run id, so none is replaced. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
irparent
added a commit
that referenced
this pull request
Oct 5, 2026
…ns never replace each other (#852) * ci: e2e keeps the Playwright browsers in a cache, and gets 40 minutes The 20-minute limit every job gained in #845 cancelled e2e on a pull request whose tests were passing: the browser download took 976 s against about a minute normally (Playwright's download server, not the change). The browsers are now cached per Playwright version, so most runs download nothing, and e2e, the one job that depends on someone else's server for a large download, gets 40 minutes; every other job keeps 20. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * ci: every run that is not a pull request gets a concurrency group of its own #845 gave ci, claims-alignment, codeql, lighthouse and publish-python one concurrency group per pull request, cancelling a run its PR replaced. On main the group was the branch, so a merge queued behind a running one replaced the merge queued before it: two of main's commits on 10-05 never got their CI run. A pull request keeps its group; every other run (a push to main, a tag, a schedule) is grouped by its own run id, so none is replaced. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What changes
Measured on main's recent CI runs (every job's p95 is under 6 minutes, and the run took 11.6 to 15.2 minutes):
lint-and-typecheck,testorbuildwithout using anything they produced.e2ewaited forbuild, then rebuilt everything itself, so it started 6–9 minutes into the run and finished last every timeneeds:, because they pass artifacts; everything else starts at onceci,claims-alignment,codeql,lighthouseandpublish-pythonkeep one run per PR; main and tags are never cancelled|| trueNothing that gates a merge changes: the same jobs run, with the same names.
Tests
preflight-mirrors-ci,workflows-build-order,release-workflow-consistency,ci-gate-action,site-says-what-the-code-does,required-checks-documented.npm run preflightpassed on this commit.🤖 Generated with Claude Code