completeness-auditlisted

# Completeness Audit A plan drifts away from 100% coverage in two silent ways. This skill catches BOTH. Materiality decides ORDER, never WHETHER an item ships (the no-defer-and-forget mandate). - **Failure (a) - soft-defer drift:** items get "parked", "on demand", "wait for a concrete failing case", "deferred opportunistic", "someday" - technically in the plan but framed optional-forever. A grep finds these (the plan mentions them). - **Failure (b) - silent source gaps:** governed content in the source docs was never catalogued in the plan at all (e.g. an entire refining method family overlooked by a note that claimed "essentially all built"). A grep CANNOT find these - the plan does not mention them. Only a source-vs-plan coverage sweep finds them. This is the more dangerous mode. ## Procedure ### STEP 1 - Soft-defer sweep (failure a) Grep the plan for the optional-forever markers (case-insensitive): `-> park`, `\bpark\b`, `on demand`, `on-demand`, `only on real user demand`, `wait for a concrete`, `deferred opportunistic`, `pick when value beats`, `someday`, `maybe later`, `if time`, `only when ... window`, `DROPPED`, `EXCLUDED`/`excluded`. For each hit, reclassify into exactly one of: - a **real SCHEDULED build item** in correct materiality order (with a home), OR - an **explicit FINALIZED justified exclusion** (state the reason; e.g. "value genuinely not in the source corpus", "process artifact, not source content"). No "optional-forever" item may remain. (The `plan