Conciseness

Mostly dense and useful, but citation rules are repeated across four sections (Contract, Output Format, Quality Rules, Citation in Answers), 'never hallucinate' appears four times, source precedence is stated three times, and the Token-Budget Awareness section carries protocol-spec internals (ceil cost estimates, fractional-budget compatibility, adoption-verification caveats) plus version/date-sensitive details ('gbrain ≥ 0.43', 'v0.10.1+', '2026-09-01') outside any deprecated section. Not 4: the redundancy and padding are more than minor; not 2: the phases, failure handling, and graph sections are tight and earn their tokens.

Actionability

Fully executable, copy-paste-ready commands cover the common cases: 'gbrain query "<q>" --since 2026-09-01 --until 2026-09-30', four worked graph-query examples ('gbrain graph-query companies/acme --type works_at --direction in'), and a recall example given in both bash and MCP JSON forms. This matches the anchor for specific examples covering common cases.

Workflow Clarity

The five-phase sequence (decompose → execute → read → synthesize → flag gaps) is explicit, phase 5 integrates notice-checking before claiming gaps, and 'When it fails' provides concrete feedback loops (degraded → try exact names/synonyms and surface the fix; page_not_found → search by title and check --source; listing_truncated → narrow before claiming absence). Clear sequence with explicit validation and error-recovery loops.

Progressive Disclosure

Good section headers, but there is no bundle: external references (skills/conventions/brain-first.md, docs/protocol/MEMORY_VERBS_v1.md, ../conventions/agent-operator-protocol.md) are buried mid-sentence in prose rather than clearly signaled, and roughly 45 lines of budget-packing protocol detail belong in a separate reference file. This matches 'some structure but content that should be separate is inline'; not 4 because the navigation to deeper material is not clearly organized.

Specificity

Lists several concrete actions — '3-layer search, synthesis, and citation propagation' — matching the anchor for several specific actions with minor gaps; graph traversal, gap flagging, and timeline lookups documented in the body are absent, keeping it below comprehensive (5).

Completeness

Explicitly answers both: what ('Answer questions using the brain's knowledge with 3-layer search, synthesis, and citation propagation') and when ('Use when the user asks a question, wants a lookup, or needs information from the brain') with concrete trigger phrases — a direct match to the anchor-5 example pattern.

Trigger Term Quality

The description's 'Use when the user asks a question, wants a lookup, or needs information from the brain' is supplemented by an explicit frontmatter triggers list ('what do we know about', 'tell me about', 'who is', 'look up', 'search for', 'background on', 'connections', 'graph query') giving comprehensive natural-term coverage including synonyms. Not 4: common variations users would actually say are already enumerated.

Distinctiveness Conflict Risk

'Using the brain's knowledge' / 'from the brain' carves a clear niche, but 'when the user asks a question' is broad enough to overlap with general Q&A or web-search skills — mostly distinct with minor overlap risk, not the minimal-conflict clarity of anchor 5.