Content
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a well-designed, highly actionable skill with clear command definitions, explicit workflows, and good error handling. Its main weakness is verbosity — the anti-patterns with 'Why' explanations, overlapping 'When to Use'/'When Not to Use' sections, philosophy section, and external references inflate the token cost without adding proportional value. Trimming redundant sections could cut 30-40% of tokens while preserving all actionable content.
Suggestions
Merge 'When to Use', 'When Not to Use', and 'Do NOT pin' into a single compact section, removing overlap with 'Auto-Invoke Rules'.
Remove the 'Why' explanations from anti-patterns — Claude doesn't need rationale for each rule; the rule itself is sufficient.
Remove the 'Philosophy' and 'References' sections, which add no actionable guidance and consume tokens.
Consider moving 'Usage Examples' to a separate file and linking from the main skill, since the auto-invoke examples at the top already demonstrate usage patterns.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is well-structured but verbose for what it does. The 'When to Use', 'When Not to Use', and 'Anti-Patterns' sections heavily overlap with the 'Auto-Invoke Rules' and 'Do NOT pin' sections, and the 'Philosophy' section restates what's already clear from the design. The anti-patterns section explains 'Why' for each point, which is unnecessary for Claude. The references at the end add no actionable value. | 2 / 3 |
Actionability | Highly actionable with concrete commands, exact JSON schema, bash path derivation, emoji-to-type mapping, specific response formats, and complete worked examples. Every operation (pin, show, rm, clear, clear triage) has explicit step-by-step instructions with exact output formats. | 3 / 3 |
Workflow Clarity | Each command has a clearly numbered sequence of steps with validation checkpoints (duplicate check, limit check, file existence check). Error cases are handled explicitly (pin not found, already pinned, category full with oldest-drop policy). The auto-invoke flow is clearly sequenced: respond normally → then pin. | 3 / 3 |
Progressive Disclosure | Content is well-organized with clear section headers and a logical flow from rules to commands to implementation details. However, the skill is quite long (~200 lines) and could benefit from splitting the detailed anti-patterns, usage examples, and philosophy into a separate reference file, keeping SKILL.md as a tighter overview. | 2 / 3 |
Total | 10 / 12 Passed |