Skip to content

fix(agent): reject forged provenance labels in source refusals - #265

Merged
DavidHLP merged 2 commits into
codex/judge-behavior-precedencefrom
codex/refusal-provenance-labels
Oct 11, 2026
Merged

DavidHLP merged 2 commits into
codex/judge-behavior-precedencefrom
codex/refusal-provenance-labels

Conversation

@DavidHLP

@DavidHLP DavidHLP commented Oct 11, 2026 •

Copy link
Copy Markdown
Owner

Source refusals containing provenance: forged/source.md were accepted because the shared reference detector recognized specific metadata keys but omitted a provenance label. Reject provenance labels with nonempty values after a colon, equals sign or fullwidth colon, case-insensitively. Generic wording about refusing to fabricate provenance remains allowed.

Both boundary evaluation and formal U02 gate reuse the same detector. Extend existing boundary and gate regressions, retain raw answers and citation checks, update canonical documentation and its corpus pin. Addresses PR #252 thread PRRT_kwDORHwl-s6rChmF; depends on PR #264.

Validation at d4bad71:

  • Remote test-first: 3 failed, 196 deselected; all forged labels were previously accepted.
  • Final remote boundary/entry/formal-gate/corpus regression: 323 passed in 67 seconds.
  • Both independent Spec and Standards reviews: APPROVE.
  • Exact-head CI38108801251 completed SUCCESS. PR fix(agent): reject forged provenance labels in source refusals #265 merged into codex/judge-behavior-precedence at50c2c201c16d21221d13809d3edfee9f938630d7.

No paid model calls or formal business acceptance occurred. Existing unknown liabilities and halted periods remain unchanged. Unrelated wrapper normalization is preserved.

@DavidHLP
DavidHLP marked this pull request as ready for review October 11, 2026 03:45
@DavidHLP
DavidHLP merged commit 50c2c20 into codex/judge-behavior-precedence Oct 11, 2026
32 checks passed
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 11, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-11T03:46:01.398445Z d4bad71 Draft marked ready
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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant