forge

Audit a CATS-style lab for drift between its stance and its artifacts (SPEC.md, SOLUTION-ARCHITECTURE.md, tests, code). Produce a gap report + a step-by-step remediation plan. Invoke when the user asks to "review the architecture", "audit drift", "do a CATS review", or after a non-trivial refactor.

Synthetic minimal skill for wiki-bench TDD test of harness capabilities (web search, prior-source reading). Run when the operator says "run the synthetic benchmark skill".