Conciseness

The body is dense and table-driven, assumes competence (never explains what IBD, EMA, or distribution days are), and every table earns its place. Minor trim candidates keep it from score 5: the Purpose paragraph restates the frontmatter description, and the Counting Conventions note referencing the Japanese phrase "直近 N 取引日" plus "Japanese explanations preserved" in Outputs expose internal i18n details irrelevant to executing the skill.

Actionability

Fully executable guidance throughout: a copy-paste-ready CLI invocation with all flags ("python3 skills/ibd-distribution-day-monitor/scripts/ibd_monitor.py --symbols QQQ,SPY --lookback-days 80 ..."), precise numeric detection rules (0.2% decline, higher volume, 25-session expiration, 5% invalidation), API-key resolution priority order, exact output filename patterns, and named pipeline functions. The common case is covered completely, matching the top anchor.

Workflow Clarity

The 9-step workflow is clearly sequenced with named functions and includes a validation checkpoint ("Validate data quality; record skipped sessions in audit") plus explicit handling of insufficient data ("flag market_below_21ema_or_50ma (None if data insufficient)"). It lacks a validate-fix-retry feedback loop (e.g., what to do when the FMP fetch fails or data quality is rejected), so it does not reach score 5; the read-only analysis workflow is not destructive/batch, so the score-3 cap does not apply.

Progressive Disclosure

The body is well sectioned and properly wired to the scripts bundle (CLI path resolves, workflow function names match real modules), but the two reference files in references/ (ibd_distribution_methodology.md, tqqq_exposure_policy.md) are never linked or signaled anywhere in SKILL.md — they are undiscoverable — and line 82 references config/default.yaml, which does not exist in the bundle. This matches "references present but not clearly signaled" (score 3) rather than score 4, whose "references mostly clear" requirement fails outright.

Specificity

Lists five concrete actions with explicit quantitative parameters: "Detect IBD-style Distribution Days for QQQ/SPY (close down at least 0.2% on higher volume)", "track 25-session expiration and 5% invalidation", "count d5/d15/d25 clusters", "classify market risk (NORMAL/CAUTION/HIGH/SEVERE)", "emit TQQQ/QQQ exposure recommendations". Coverage is comprehensive with no minor gaps, matching the top anchor rather than score 4.

Completeness

Clearly answers both questions: the "what" is detection, expiration/invalidation tracking, cluster counting, risk classification, and exposure recommendations; the "when" is explicit with three concrete triggers — "Use after market close, before TQQQ exposure changes, or as input to FTD/market-state frameworks" — plus a negative boundary ("Does not execute trades"). This matches the top anchor; the when-clause is explicit and specific, not merely present as at score 4.

Trigger Term Quality

Good natural keyword coverage: "Distribution Days", "QQQ/SPY", "TQQQ exposure", "market risk", "after market close", "FTD". A few natural phrasings a user might say are missing (e.g., "market top", "correction risk", "institutional selling", "reduce exposure"), so it falls short of the comprehensive synonym coverage of score 5, but is well above the sparse keyword coverage of score 3.

Distinctiveness Conflict Risk

The niche is distinct ("IBD-style Distribution Days", "d5/d15/d25", QQQ-weighted logic) and the primary trigger "distribution day" routes here unambiguously, but "classify market risk" and "emit TQQQ/QQQ exposure recommendations" overlap with closely related sibling skills named in the body (market-top-detector also uses O'Neil distribution; position-sizer consumes exposure recommendations). This is exactly "mostly distinct; minor overlap risk with closely related skills", not the minimal-conflict profile of score 5.