Conciseness

The body is dense operational guidance — tool calls with a literal `appendResearchLog` JSON payload, call limits, merge rules ("lowercase, trimmed, collapsed whitespace"), registrable-domain counting, and cost checkpoints — with no explanation of concepts Claude already knows; the provenance caveat repeated in the report section ("How this report was made") is required disclosure, not padding. This fits the 5 anchor (lean, assumes competence, every token earns its place) rather than 4, which requires over-explanation that could be trimmed.

Actionability

Exact tool names (`research_ai_visibility_prompts`, `explore_prompt`, `get_search_console_performance`, `update_project_context`), concrete quantities (5–15 head terms, two to four chosen, 10–20 prompts to track), a copy-paste-ready log payload, and explicit fallbacks ("If the call fails, such as no paid plan, skip it and say so") make the guidance fully executable for an instruction-only skill. Matches the 5 anchor; the minor-gap level of 4 does not apply.

Workflow Clarity

A numbered five-step workflow with explicit validation checkpoints and feedback loops: reuse research log under 30 days before paying again, state the market limitation "before spending", retry once with a broader term on empty results, ask before spending credit on `explore_prompt`, skip-and-say on failure, plus a section-by-section deliverable checklist. Fits the 5 anchor; step 4's sequence has no meaningful validation gap to drop it to 4.

Progressive Disclosure

The single file is well organized (Goal, Project context, Tools, Workflow, What to return, Guardrails) with clearly signaled cross-skill references (`seo-report`, `ai-visibility-audit`, `keyword-research`), but at roughly 90 lines it exceeds the under-50-line simple-skill exception, and reference material such as the MCP tool details and the report template could be split into one-level-deep files. Good structure with minor organization headroom fits 4 rather than 5.

Specificity

"Find the questions people ask about a market, how ChatGPT answers them, and which sites get cited in the answers" lists three concrete actions covering the full pipeline, plus an explicit boundary ("Research only; it never saves tracking or starts answer collection"), in third-person voice. This matches the 5 anchor (multiple specific concrete actions, comprehensive coverage) rather than 4, since no capability area is missing.

Completeness

A clear "what" (find questions, how ChatGPT answers them, which sites get cited) is paired with an explicit "Use when" clause containing three concrete trigger phrases ("asks what people ask AI about their niche, wants prompt ideas, or wants to know where their brand is missing"). Clearly the 5 anchor; not 4 because the "when" is explicit and specific.

Trigger Term Quality

Natural user phrasings are present ("what people ask AI about their niche", "prompt ideas", "where their brand is missing", "prompts to track"), but common variations such as "AI visibility", "AI search", "ChatGPT recommendations", or "citations" are absent. Good coverage with a few natural terms missing fits the 4 anchor, not the comprehensive-synonyms bar of 5.

Distinctiveness Conflict Risk

The niche is clear (AI-assistant question research and citation analysis) and it explicitly disambiguates from Prompt Tracking ("Research only; it never saves tracking or starts answer collection"), but "wants prompt ideas" and "where their brand is missing" retain minor overlap risk with closely related skills such as keyword research or an AI-visibility audit. Mostly distinct with minor overlap fits 4, not the minimal-conflict bar of 5.