experiment: seal critical PTCNA evaluation result

[erinepshovel-code]

Frozen result

• usefulness: FALSIFIED (target 0.3333333333 < 0.75)
• superiority: FALSIFIED (target 0.3333333333 vs fallback 0.9444444444; advantage -0.6111111111 < 0.05)
• 270 training steps and 90 evaluations completed under the 120-second limit
• no backend or resource failure
• plan digest: 67cdad3aefb3e33f6fbf3994de54e1b73a01105527bf241fce08947ed7046bbe
• result digest: 3d73f08f3e0eeaf0f5e508aba762844c9e8d11b7146b654fd562aff425e30f13

The result was recorded before repair. Construction compatibility remains established; generalization, continuous geometry, EDCM validity, external validity, and production privacy remain unestablished.

Validation

• independent arithmetic replay agrees with both verdicts
• 171 tests
• contract graph CLOSED
• ratios 39/39, 234 verified, zero drift/gaps
• deterministic metadata collection
• sdist/wheel + Twine + isolated wheel receipt smoke

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5c24444678

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".

[chatgpt-codex-connector]

Record the commit that actually generated the receipt

When an auditor follows source_commit, this hash checks out the pre-preregistration tree: it lacks the frozen plan/runner and its evaluation.py still uses the earlier single-status/deficit API, so it cannot generate or validate this receipt. The documentation says execution occurred only after f202800... was merged; preserve the 3d67f... backend baseline separately and record the actual execution/evaluator commit in the sealed JSON so the frozen metrics and failure semantics are reproducible.

AGENTS.md reference: AGENTS.md:L35-L38

Useful? React with 👍 / 👎.