Conciseness

The body is dense with operational content — commands, thresholds, a domain-shape table, anti-patterns — with only minor trimmable framing such as the 'The Bug This Kills' anecdotes and the rhetorical 'The Standard' closer. Efficient with minor instances of over-explanation (anchor 4), not 5 (the motivational sections don't strictly earn their tokens) and not 3 (there is no padding that teaches concepts Claude already knows).

Actionability

Fully executable throughout: copy-paste-ready commands ('gbrain think "Who is {entity}?..."', 'gbrain search "{entity}" --limit 5', 'gbrain timeline {entity-slug} --limit 20'), exact placeholder markers to scan for ('[To be filled', 'Unknown relationship', 'TBD'), concrete batch parameters ('batches of 10–20', '100+ pages'), and a complete escalation output template — specific examples cover the common cases.

Workflow Clarity

The six-step chain is explicitly ordered with 'STOP at the first clear answer' validation checkpoints, confidence thresholds define the stop/ask decision, the post-ingest batch pass opens with a validation/skip check (step 1 placeholder scan) plus a quality-bar acceptance test, and 'When it fails' provides a feedback loop ('revision_conflict: re-read and merge'). The batch-operation cap does not apply because validation and recovery steps are present.

Progressive Disclosure

Thirteen clearly headed sections with sibling-skill references (../conventions/brain-first.md, ../ask-user/SKILL.md, ../query/SKILL.md, etc.) that are one level deep and well-signaled via the top convention blockquotes and the sharp-boundary Dedup section. However, no bundle files exist (no references/, scripts/, or assets/), so all ~290 lines are inline — the ingest batch mechanics and domain-shortcut table could live in a separate reference file — which keeps it below the top anchor and outside the under-50-line simple-skill exception.

Specificity

The description enumerates concrete actions: 'exhaust the brain's lookup chain: think → search + page read → mounted sources → timeline/graph → web', 'ask WITH a hypothesis, never a bare unknown', and 'pages created during bulk ingestion get their relationship/role resolved immediately' — comprehensive coverage of the skill's capabilities, matching the top anchor rather than the 'minor gaps' of a 4.

Completeness

The 'what' is explicit and multi-part (gate identity questions, run the lookup chain, escalate with a hypothesis, resolve placeholders at ingest), and a 'when' is present via 'Before any "who is X?" (or role / relationship question) reaches the user' — an equivalent trigger condition that clears the 3-cap. It is not a 5 because the explicit 'Use when...' phrasing with concrete trigger phrases lives in the separate when_to_use field rather than in the description itself.

Trigger Term Quality

Natural phrases like "who is X?", 'role / relationship question', 'placeholders', and 'bulk ingestion' are present, but common user variations ('who is this person', 'unidentified contact', 'unknown') appear only in the separate triggers field, not the description. Good coverage with a few natural terms missing — anchor 4, not 3 (keywords are relevant and naturally phrased, not generic) and not 5 (synonyms and variations are not included in the description text itself).

Distinctiveness Conflict Risk

'Gate on identity questions to the user' and the 'no-placeholders-at-ingest rule' define a clear, narrow niche — a decision gate on one class of question plus an ingest acceptance bar — with trigger conditions unique to this skill, so conflict risk with related lookup/asking skills is minimal.