docs: document the claim review flow and add a manual smoke test checklist#72
docs: document the claim review flow and add a manual smoke test checklist#72enlorik wants to merge 2 commits into
Conversation
|
@codex review |
![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: 2563747076
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".
| const SizedBox(height: 2), | ||
| Text( | ||
| key: const Key('claim-review-provider-label'), | ||
| 'Answered via $providerName', |
There was a problem hiding this comment.
If a grounded-answer provider accidentally sets GroundedAnswer.providerName to a credential-like value (the ingestion layer does not sanitize or validate it), this label renders that raw string directly on screen, contradicting the documented/agent rule that API keys and provider credentials are never visible in the UI or screenshots. Use a fixed safe display label from the provider configuration or redact credential patterns before interpolation.
Useful? React with 👍 / 👎.
|
@codex review |
|
Codex Review: Didn't find any major issues. Another round soon, please! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
163070e to
11c38a5
Compare
1 participant
Realized there was nowhere for anyone (including future me) to figure out how the claim review flow actually works or how to poke at it manually, so I wrote it up. Added a "Claim review" section to the README covering the ask -> review -> draft -> save/append flow and the safety rules (no auto-save, selected-only, already-known never saved, no fake Brave AI Answers, no leaked keys), plus a step-by-step manual smoke checklist in docs/claim_review_smoke_test.md that points at the existing end-to-end test suite for the automated equivalent. Docs only, no code changes — 516 tests still passing, analyze still clean.