Content
56%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.
The skill body delivers concrete, executable repomix commands and a coherent four-step workflow with error-recovery guidance, but it is noticeably padded and monolithic. Roughly half the file (intent examples, communication style, generic grep recipes, self-verification checklist) duplicates what Claude already knows and should be trimmed or moved to reference files.
Suggestions
Cut the 'User Intent Examples', 'Communication Style', and 'Self-Verification Checklist' sections — they restate the description's triggers and generic behavior Claude already follows, and they drive the conciseness score down.
Move the 'Search Patterns' grep catalog and the five 'Example Workflows' into a single references/ file (e.g. references/patterns.md) and link to it from a short 'Workflow' section, so SKILL.md stays a lean overview.
Add an explicit validation checkpoint in Step 1/Step 2: confirm the repomix command's exit status and that the output file exists (e.g. test -f /tmp/<repo>-analysis.xml) before grepping or reading it, upgrading the workflow from implicit to explicit validation.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Several padded sections restate what Claude already knows: the role preamble 'You are an expert code analyst...', ~20 'User Intent Examples' that duplicate the description's triggers, 'Your Responsibilities' items like 'Understand the user's intent from natural language', a 'Communication Style' section advising 'Be concise but comprehensive' and 'Cite sources', generic grep recipes ('grep -iE "error|exception|try.*catch"'), and an 11-item 'Self-Verification Checklist' ('Did you communicate clearly and concisely?'). This matches the 'noticeably verbose; several unnecessary explanations or padded sections' anchor; not 1 because the core command/flag material is genuinely useful and not conceptual filler. | 2 / 5 |
Actionability | Concrete, executable commands throughout: 'npx repomix@latest --remote yamadashy/repomix --output /tmp/repomix-analysis.xml', a flags table ('--compress', '--include "**/*.{ts,tsx}"', '--remote-branch'), and runnable greps. Not 5 because some guidance remains narrative (Example Workflow steps like 'Summarize: "This repository contains [number] files..."' and 'Provide structural overview with ASCII diagram if helpful') rather than copy-paste-ready instructions, leaving minor gaps. | 4 / 5 |
Workflow Clarity | A clear four-step sequence (Pack the Repository -> Check Command Output -> Analyze the Output File -> Provide Insights) with an Error Handling section providing feedback loops ('Command fails' -> verify URL/path/permissions; 'Pattern not found' -> 'Try alternative patterns', 'Suggest broader search'). Not 5 because there is no explicit validation that the pack command actually succeeded (exit status / output file exists) before proceeding to grep, and the Self-Verification Checklist is stylistic self-assessment rather than an operational checkpoint; not 3 because recovery loops and a mid-workflow output check are present. | 4 / 5 |
Progressive Disclosure | The body has good section headers but is a single ~295-line monolith with no references/ or other bundle files at all: content that clearly belongs in separate files is inlined, such as the 'Search Patterns' recipe catalog, five full 'Example Workflows', and the long intent-example lists. This matches the anchor 'content that should be separate is inline' while structure is present; not 4 because essentially all bulk detail lives in SKILL.md rather than being offloaded to well-signaled reference files. | 3 / 5 |
Total | 13 / 20 Passed |