auditlisted

# Audit the evidence trail Read the project records and inspect the files. Do not change scientific content during the audit. ## Checks Check every item in these groups: 1. Find manuscript claims that do not cite experiment evidence. 2. Find experiments with no pre-run expectation, result, interpretation, insight, or hypothesis link. 3. Find figures whose source experiment and output are unknown. 4. Find superseded experiments that an insight, figure, or claim still cites. 5. Find loose files at the root or files stored outside their fixed place in [AGENTS.md](../../../AGENTS.md). Also check that Markdown links resolve, experiment numbers are unique, index rows match folders, and frozen code and configuration exist. ## Report Write a plain-language report. Put problems that can invalidate a claim first. For each problem, name the file, explain the effect in one sentence, and offer a one-line fix. Separate confirmed problems from items that need human judgment. End with no more than three scientific questions that only the researcher can judge. If there are no such questions, state that clearly. Do not create extra questions to fill the limit.