Audit Paper

Audit a paper's empirical claims against the linked code repository — catches MISMATCH, MISSING, and PARTIAL claims before submission

Audit Paper vs. Code: $ARGUMENTS

Activate the audit-engine skill. Read skills/audit-engine/SKILL.md in full.

Execute the complete 5-step audit workflow:

  1. Extract testable claims from paper.tex (or draft.md)
  2. Locate the code repository (from $ARGUMENTS, or scan paper for a code URL, or ask once)
  3. Map the repo (entry points, configs, data loaders, results)
  4. Search for evidence per claim using Grep + Read with file:line citations
  5. Classify each claim as CONFIRMED / PARTIAL / MISSING / MISMATCH / NOT_AUDITABLE
  6. Generate outputs/audit_report.md

$ARGUMENTS may contain:

  • A repo path or GitHub URL to audit against
  • A scope hint (e.g., "only Section 4", "only the training claims", "tier 1 only")
  • Both

If no code repository can be located, stop and ask the user for a path or URL before proceeding.

This is a static audit: no code is executed. For full reproducibility, re-run training and evaluation manually after the audit surfaces any MISMATCH or MISSING claims.