The body is mostly efficient and avoids explaining concepts Claude already knows, but Step 0's seven per-CMS quoted setup dialogs and repeated confirm phrasings ('Want me to continue...?') could be tightened, and lines like 'This saves the user from a copy-paste dead end at the finish line' add mild padding. It is not maximally lean, so it does not reach the score-3 anchor.

Provides concrete tool names (list_ai_visibility_org_brands, get_ai_visibility_topics, wp_create_post, create_entry), explicit field mappings (fields.title, status: draft, state: DRAFT), quantified heuristics (relevancy > 60% and visibility < 40%), and character limits. Though instruction-only with no code, the guidance is specific and copy-paste ready, which the rubric rewards.

Steps 0–6 are clearly sequenced with explicit confirmation checkpoints at the risky publish step ('Ready to push this to [CMS] as a draft?') and a guard ('never published without explicit instruction'). The sequence and checkpoints are explicit rather than implied, placing it above the score-2 anchor.

Sections are well-organized with clear headers and tables, but the skill is a monolithic ~250-line single file with no bundle references, and the lengthy per-CMS setup details in Step 0 and field mappings in Step 6 are content that could plausibly live in a separate reference file. It is well structured rather than a wall of text, so it is above 1, but lacks the one-level-deep references needed for a 3.

Lists multiple concrete actions — 'turn AI Visibility data into published content', 'produces a publish-ready content brief plus full article draft', and 'push the draft directly' to a CMS — alongside specific data types (topics, prompts, visibility scores, citations). It is comprehensive, not merely naming a domain, so it sits above the score-2 anchor.

Explicitly answers both 'what' (connects to Amplitude AI Visibility data, produces a brief plus draft, pushes to CMS) and 'when' ('Use this skill whenever a user wants... whether they say...'), with explicit trigger guidance, so it is not capped at 2.

Quotes many natural phrasings a user would actually say ('find content gaps', 'what should we write about', 'help me get cited by AI models', 'what content should we create?'), giving strong coverage of common variations rather than a single keyword.

The niche is narrow and specific (Amplitude AI Visibility gaps → published content/CMS draft) with triggers tied to an AI Visibility context, making conflict with unrelated skills unlikely.