Implements §1 of the parent design — rung 0 of Thread B, and the gate on whether Thread B has a mechanism worth building at all.

Copy the frozen goldv2-20260812 corpus. Hand-write ideal aliases: frontmatter, in question vocabulary, for exactly the eight unreachable notes (pp05, pp07, pp09, pp15, pp16, pp17, rc01, rd01). All eight are already verified admitted by recall's hygiene filter, which places their failure in vocabulary. Rebuild a scratch index, assert its integrity, and re-score the +question arm and the hook-shaped path.

The oracle answers one question: if the alias were perfect, would the note be found? A hand-written ideal alias is the most favourable input any alias engine could ever produce, so this bounds the whole thread from above before a line of engine code exists. It mirrors the term-subset oracle that bounded the lexical thread at 82.8% in the previous arc.

None. This part is deliberately first: it is deterministic, needs no LLM, runs in an afternoon, and its result decides whether alias-pilot is built or closed unbuilt.

• At least 6 of the 8 target questions convert to top-5 hits.
• No currently-passing question is lost. The per-question gain/loss diff is published by id — added alias text changes both BM25 statistics and dense vectors, and reciprocal-rank displacement has cost this project hits three times across the previous arc.
• No stratum regresses by more than one question.
• Index integrity asserted before scoring: 9,971 docs, and a file-level diff against the archived snapshot showing exactly the eight edited notes differ.
• Nothing derived from this oracle ships. It is gold-informed by construction, under the same diagnostic licence as the candidacy analysis and the k=20 reachability count.

If the rule fails (<6/8): the part closes as refuted, recorded with the numbers. The misses are then not vocabulary-shaped, alias-pilot closes unbuilt, and the recall clause of the arc-close gate is re-priced before any money is spent on it.

This part implements one slice of AgentM Rejection and Vocabulary (Status: final). See the parent for Context, Alternatives Considered, the Quality Attributes overview, and the Operations strategy — in particular the gold-blindness boundary in §1, which governs this part and every part downstream of it. Mid-execution changes to this part's scope must be appended to the parent's Document History.