Conciseness

The body is dense policy rather than padded explanation, but there is structural redundancy: "Outputs" and "Output Format" list overlapping field sets, and "Gotchas", "Anti-Patterns", and "Constraints" restate the same guardrails three different ways. It fits anchor 3 (mostly efficient but could be tightened) rather than 4, because the duplication is structural across sections, not minor trims.

Actionability

Concrete anchors exist — artifact routes (`.harness/brainstorm/**.md`), the validation command (`python3 Plugins/harness-engineering/scripts/check_bluf_structure.py ... --json`), and explicit output field names — but core behaviors are invoked by name only ("apply the BLUF review contract", "apply the first-principles contract", "apply the stage arc boundary") with no inline summary, so execution requires chasing external files. This matches anchor 3 (some concrete guidance but incomplete, missing key details) rather than 4, where gaps would be minor.

Workflow Clarity

A 12-step sequenced procedure, an explicit Validation section with pass/fail/blocked statuses and a runnable script, and Failure Handling with recovery loops ("stop with the blocker and smallest recovery step", "ask once; otherwise set `autonomous_assumption`") give clear checkpoints with feedback loops. It fits anchor 4 rather than 5 because validation is conditional ("non-trivial generated artifacts") and is a separate section rather than an explicit step wired into the numbered sequence.

Progressive Disclosure

References are clearly signaled with "Read when..." conditions, but the actual bundle structure reveals problems: local reference files chain onward to `Infrastructure/references/harness-engineering/he-brainstorm-doctrine.md` (2+ levels of nesting), roughly ten `../../references/*.md` targets sit outside the bundle entirely, and `references/hot-path-folded-context.md` duplicates body content (Procedure, Examples, References). This lands at anchor 3 — better signaled than anchor 2's buried references, but the nested chains and duplication exceed anchor 4's "minor organization gaps".

Specificity

The description lists four concrete actions ("Explore Harness Engineering options, filter ornate or weak ideas, recover dropped leverage, and select survivor routes before commitment"), but insider jargon such as "recover dropped leverage" and "select survivor routes" leaves minor gaps in how comprehensible the actions are. It exceeds anchor 3 (only 1-2 actions) but falls short of anchor 5 because the actions are not fully self-explanatory without domain context.

Completeness

It explicitly answers both what ("Explore... options, filter ornate or weak ideas, recover dropped leverage, and select survivor routes") and when ("Use when intent, stage choice, tradeoffs, idea quality, or possible solution shapes are still unsettled before spec, plan, Linear, or implementation work") with concrete trigger conditions. This matches anchor 5 exactly; the 'when' clause is explicit, not merely implied, so 4 does not fit.

Trigger Term Quality

Natural terms like "options", "ideas", "tradeoffs", "spec", "plan", and "intent" are present, but the single most natural user term for a brainstorm skill — "brainstorm" — never appears, nor do common phrasings like "help me decide" or "weigh options". This matches anchor 3 (relevant keywords but missing common variations) rather than 4, since the missing head term is a variation users would actually say.

Distinctiveness Conflict Risk

"Harness Engineering" branding plus the explicit pre-commitment scope ("before spec, plan, Linear, or implementation work") carves a clear niche distinct from downstream stages. It is not 5 because generic words ("options", "tradeoffs", "spec", "plan") and proximity to sibling he-ideate/he-spec skills leave minor overlap risk, but it is well above anchor 3's generic overlap.