RAG canary: distinguish content, provider, and latency variability#1095
Conversation
Persist structured answer-quality reports with run context and add stable content, provider, and latency comparison fingerprints. No retrieval or answer behaviour changes.
|
This pull request has been ignored for the connected project Preview Branches by Supabase. |
|
Warning Review limit reached
Next review available in: 28 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (7)
✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3e3831be4a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
@codex resolve actionable Codex review findings for this pull request and current head using the repository instructions. This is the pull request's single automatic repair pass: do not perform a fresh review, create new standalone findings, or request another review. Work only the existing unresolved Codex threads on the current head. Always fix P0 and P1 findings. For P2 and lower findings, fix only clear, scoped, low-risk issues; otherwise disposition them with a concise reason. After fixing or dispositioning a thread, reply in that thread with as the first line, followed by a concise summary; that marker authorizes the workflow to close that exact thread. If human input or new authorization is required, do not use the marker and leave the thread open with the blocker. Finish only after every actionable thread is fixed or dispositioned and closed, or explicitly left open for a human decision. Do not update the branch from main, address unrelated reviews, broaden scope, or create more than one scoped fix commit. Do not use external APIs, paid services, credentials, dependency changes, or broad refactors unless explicitly authorized. Add targeted tests where behavior changes and run the narrowest relevant validation. |
Export the checked-out SHA so optional-ref canary artifacts cannot be mislabeled as same-tree comparisons.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3e3831be4a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
1 participant
Summary
Why
A byte-identical main tree produced different answer-quality outcomes across live runs. The existing artifact retained logs but not enough stable per-case evidence to distinguish nondeterminism from a deterministic content regression.
Verification
npx vitest run tests/eval-quality.test.ts tests/eval-trend.test.ts tests/eval-canary-workflow.test.ts --reporter=dot— 38/38 passednpm run check:github-actions— passednpm run eval:rag:offline— 21 suites, 307/307 passed; 36 golden cases validatednpm run check:rag:fixtures— passednpm run verify:cheap— 365 files, 3,235 passed, 1 skipped; lint and typecheck passedgit diff --check— passednpm run verify:pr-local -- --dry-run --files ...— selection inspected; full duplicate run not repeated becauseverify:cheapplus the focused RAG checks covered the selected commandsRisk and rollback
Diagnostics/workflow-artifact change only. The next live run must validate the report content before #51 is closed. Rollback is a plain revert of this commit.
RAG impact: no retrieval behaviour change — only eval report provenance, offline comparison diagnostics, tests, workflow artifact paths, and documentation changed.
Checks not run
Note
Low Risk
Eval diagnostics, workflow artifact paths, and documentation only—no retrieval or answer-pipeline behavior changes; weekly canary cost rises from 8 to 44 generated answers per run.
Overview
Eval canary now writes structured answer-quality JSON/Markdown under
.local/eval-canary/quality-reports(uploaded witheval-canary-output), and runbook/docs align the weekly answer gate with the full 44-case committed set (was documented as 8).Report provenance:
eval:qualityJSON includesrun_context(git_sha, GitHub run id/attempt,latency_contextfrom env) so downloaded artifacts tie outcomes to the tested tree and runner context.Offline comparison:
npm run eval:trend -- --answer-qualitycompares structured quality reports using stable per-case fingerprints that split content/citation outcomes, provider/route, and latency-threshold signals. Same-SHA divergent content is labeledsame_tree_content_variabilityinstead of being read as a code regression; optional--casedrills into route, latency, and diagnostic signatures.Reviewed by Cursor Bugbot for commit 3e3831b. Configure here.