verify-claims

# Verify Claims Take a report — a subagent's, a lane's, a plan's, or your own from earlier — and re-derive its load-bearing claims instead of trusting how they sound. Full doctrine: `docs/agent-craft.md` §4. ## Procedure 1. **Extract the load-bearing claims** — those whose falseness would change the decision being made. Ignore decoration; auditing everything dilutes the audit. 2. **Re-derive each from a primary source.** Source hierarchy: running code > tests > docs > comments > memory. Each step down the ladder is a step toward hearsay. 3. **Use a different path than the claim arrived by.** Claim from reading code → check with a runtime observation. Claim from a test → read the code the test exercises. Two derivations sharing a path share that path's blind spot. 4. **Verify decision-material measurements against primary evidence.** Recompute derived counts from raw records. For test, coverage, and timing claims, inspect the actual command/workload, terminal output, and relevant source/environment basis; a summary or inherited counter is insufficient. Match a named commit/artifact exactly; for dirty work, inspect the relevant current paths and account for concurrent changes. Apply root **Validation** to reuse versus re-execution: rerun when evidence is missing, stale, contradictory, cannot establish the claimed result, or independent risk-selected verification requires it. A new handoff alone does not require repeating every successful suite. Decorative counts should be re...