The body assumes Claude's competence (no explanations of what npm/PyPI/MCP are) and uses efficient lookup tables, but repeats the frontmatter intro and re-states the Trigger list in a separate 'When to Use' section, so it could be tightened.

Provides concrete, runnable guidance: named packages per category, exact search queries, a fill-in subagent prompt template, a signal-to-action decision matrix, and worked examples with specific packages and scores. As an instruction skill this meets the actionable bar.

A clearly sequenced 5-step pipeline (Need → Parallel Search → Evaluate → Decide → Approval Checkpoint/Implement) with an explicit approval gate before any write action and a decision matrix that branches on evaluation signals.

No bundle files exist, so there are no nested references; sections are well-organized and single-level. But the skill exceeds the simple-skill line count and keeps long examples and category shortcut tables inline rather than splitting them into reference files.

Names the domain (existing tools, libraries, patterns) and one concrete action ('Search... before writing custom code'), but does not enumerate multiple distinct actions, so it falls short of the multi-action level-3 anchor.

Explicitly answers both 'what' (research-before-coding: search for existing solutions) and 'when' ('Use when starting new features or adding functionality'), matching the level-3 anchor with an explicit Use-when clause.

Includes natural triggers ('Use when starting new features or adding functionality'), but coverage is narrow and misses common variations a user would actually say ('add X', 'is there a library for', 'build a utility').

The search-before-build framing is a recognizable niche, but the 'starting new features or adding functionality' trigger is broad enough to overlap with general planning or architecture skills.