Conciseness

The ~930-line body extensively explains concepts Claude already knows (how to formulate a web search, evaluate source credibility, navigate documentation, read a README) and pads every operation with repeated 'When to Use / Prerequisites / Steps / Expected Outcome / Validation / Common Issues / Tips' scaffolding. The credibility hierarchy and search-query advice appear multiple times. This is noticeably below the midpoint — most of the per-operation prose could be cut without losing actionable content — though it stops short of anchor 1's purely elementary tutorial tone.

Actionability

There is concrete material — example search queries ('"MCP server" database access', 'topic:claude-code language:markdown'), a copy-paste synthesis template, and example findings documents — but much of the step guidance restates what Claude trivially knows ('Use WebSearch tool with formulated query', 'Review README for relevance', 'Check stars, activity'). Compared to anchor 4, the guidance is incomplete: no executable code in the body, generic process directions where specific non-obvious instructions are needed, though it is not merely vague as in anchors 1-2.

Workflow Clarity

The five operations are clearly introduced up front ('Task-based (5 independent research operations)'), each is sequenced with numbered steps, and every operation ends with an explicit validation checklist. 'Common Issues' with paired Solutions provide the error-recovery feedback loops anchor 5 requires (e.g. 'Search too broad → add specific technical terms', 'Conflicting information → prioritize official docs'), and the Research Quality Checklist closes the overall process. It clearly exceeds anchor 4's 'most checkpoints present'.

Progressive Disclosure

The bundle is real and well-matched — four references/ guides (one per research type) plus scripts/research-helper.py, all one level deep — but the body only gestures at them generically at the very end ('see the references/ directory') rather than signaling them per section, and ~600 lines of per-operation detail that duplicates the reference files are fully inlined in SKILL.md itself. This matches anchor 3 ('references present but not clearly signaled; content that should be separate is inline'); the good section structure keeps it from anchor 2, while the duplication and weak signaling keep it from anchor 4.

Specificity

The description lists several specific actions — 'discovering patterns, best practices, and technical knowledge', 'researching technologies', 'exploring codebases', 'finding examples', 'gathering requirements' — across named sources (Web search, MCP servers, GitHub repositories, documentation). It falls short of anchor 5 because the verbs remain somewhat generic ('researching', 'gathering') and it omits concrete operations like synthesizing findings or comparing MCP servers, while 'Comprehensive research toolkit' is mild padding.

Completeness

It explicitly answers 'what' (discover patterns, best practices, and technical knowledge across Web search, MCP servers, GitHub, documentation) and 'when' with a concrete 'Use when...' clause listing four trigger scenarios. Both halves are explicit with concrete trigger phrases, matching anchor 5; anchor 4's 'when could be more explicit' does not apply.

Trigger Term Quality

The 'Use when' clause provides good natural trigger phrases ('researching technologies', 'exploring codebases', 'finding examples', 'gathering requirements for skill development'). It misses common variations users might say, such as 'web search', 'look up', 'best practices', 'find documentation', so it does not reach anchor 5's comprehensive synonym coverage.

Distinctiveness Conflict Risk

The enumeration of specific channels (Web search, MCP, GitHub, docs) and the skill-development framing give it a recognizable niche with distinct triggers, but 'researching technologies' is broad enough to overlap with general web-search or code-exploration skills, keeping it below anchor 5.