[codex] Add challenge debrief feedback release guard

[AlonePenguin]

/claim #18

Summary

Adds a dependency-free challenge-debrief-feedback-release-guard module for the Scientific Bounty System. The guard evaluates synthetic post-evaluation solver debrief packets before they are released to winner, finalist, and non-winner teams.

It produces deterministic release decisions:

• release_debrief
• release_with_monitoring
• revise_before_release
• hold_debrief_release

What It Checks

• Equal rubric-linked feedback coverage across solver teams.
• Actionable, rubric-tied comments for every accepted criterion.
• Appeal deadline consistency with the challenge policy.
• Reviewer anonymity leaks.
• Private sponsor benchmark and holdout label leaks.
• Other team IP or protected method leaks.
• Civility and defamation risk in solver-facing language.
• Prohibited direct-payment, payout-negotiation, or off-platform payment language.
• Unknown team and release-target routing.

Non-Overlap With Existing #18 Slices

This PR is scoped to the final feedback packet itself. It intentionally avoids the already-covered lanes for intake compliance, rubric readiness, scoring, arbitration, reviewer consensus, award transparency, appeals handling, milestone progress, clarification freeze, evidence freeze, source IP redaction, payout eligibility, payout routing, escrow readiness, data-room access, benchmark leakage during evaluation, deadline fairness, and sponsor regulatory attestation.

Artifacts

• challenge-debrief-feedback-release-guard/index.js: evaluator and Markdown/SVG renderers.
• challenge-debrief-feedback-release-guard/sample-data.js: clean, revision-needed, and release-hold synthetic packets.
• challenge-debrief-feedback-release-guard/test.js: dependency-free assertions for the release, revise, and hold paths.
• challenge-debrief-feedback-release-guard/reports/debrief-release-packet.json: generated evaluation output.
• challenge-debrief-feedback-release-guard/reports/debrief-release-report.md: generated reviewer report.
• challenge-debrief-feedback-release-guard/reports/summary.svg: generated visual summary.
• challenge-debrief-feedback-release-guard/reports/demo.mp4: generated 8-second demo video.

Validation

Run from challenge-debrief-feedback-release-guard:

• npm run check
• npm test
• npm run demo
• npm run verify-video

Additional local safety check:

• Searched for credential-like material and user payment handles; no matches were found. Payment-related strings in this PR are synthetic detector fixtures only, used to verify that unsafe debrief language is blocked.

Data and Access

All fixtures are synthetic. This PR does not use credentials, live solver submissions, private challenge data, payment accounts, webhooks, or external APIs.

AI-assisted implementation prepared by Codex for AlonePenguin.