Content
75%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-built instructional skill: parameterized tool invocations, a concrete report skeleton, prompt-injection defenses, and a clear six-step workflow. It is held out of the top band by redundant trigger sections, a non-literal subagent snippet, the absence of explicit validation checkpoints, and a report template that could be split into a reference file.
Suggestions
Cut or shrink the 'When to Activate' and 'Examples' sections — both restate trigger phrasing already present in the frontmatter description, saving tokens without losing information (conciseness).
Add an explicit validation checkpoint in the workflow, e.g. 'Before writing, verify every cited URL was actually fetched and flag any sub-question with fewer than 2 independent sources' with a fix/retry loop (workflow_clarity).
Replace the prose subagent snippet with a literal Task-tool invocation example (prompt text, tool name, expected return format) so the parallel pattern is copy-paste executable (actionability).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is instruction-focused with no explanations of concepts Claude already knows, but minor trimming is possible: the 'When to Activate' section largely restates the frontmatter description, and the 'Examples' section repeats trigger phrasings already given. Fits 'Efficient; minor instances of over-explanation that could be trimmed' rather than the lean level-5 anchor. | 4 / 5 |
Actionability | Concrete, parameterized tool calls ('firecrawl_search(query: ..., limit: 8)', 'web_search_advanced_exa(... startPublishedDate: "2025-01-01")') and a full report template make the guidance mostly executable. It falls short of level 5 only because the subagent fan-out snippet ('Launch 3 research agents in parallel: ...') is a directive sketch rather than literal invocations, and the flagged drift-prone tool names mean some calls may not resolve as written. | 4 / 5 |
Workflow Clarity | A clearly sequenced 6-step workflow with an explicit fallback ('If the user says "just research it" — skip ahead') and Quality Rules acting as implicit checks. No explicit validate-then-fix checkpoint exists (e.g., confirming citations resolve or target source count was met before writing), which keeps it at the level-4 'minor validation gaps' anchor rather than level 5. | 4 / 5 |
Progressive Disclosure | No bundle files exist, and the single-file body is organized into well-labeled, navigable sections with nothing buried. It sits at 'Good structure; most content is appropriately placed; minor organization gaps': the ~30-line report template and the MCP configuration notes could plausibly live in one-level-deep reference files, though inlining them is defensible for a skill this size. | 4 / 5 |
Total | 16 / 20 Passed |