Content
78%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 body is a well-organized, mostly lean single-purpose skill with concrete tool invocation and a conditional fallback path. Its main gap is the fallback's placeholder path requiring derivation rather than being copy-paste ready.
Suggestions
Make the fallback path concrete (e.g., resolve the plugin root inline with `$(dirname ...)`) so the bash snippet is copy-paste executable without manual substitution.
Tighten the cross-client note to the essential "renderer-safe plain-text prefixes" detail, dropping the example client parenthetical unless it is load-bearing.
Add a brief one-line verification that diagnostics actually ran (e.g., confirm a non-empty status report was returned) to strengthen workflow clarity.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence — concrete tool name, verbatim-display instruction, and a tight fallback — with only minor explanatory clauses (e.g., the Z.ai GLM cross-client note) that could be trimmed but earn their place as operational specifics. | 4 / 5 |
Actionability | Gives the concrete `ctx_doctor` MCP tool call and an executable fallback bash one-liner, but the fallback uses a `<PLUGIN_ROOT>` placeholder requiring path derivation rather than being fully copy-paste ready. | 4 / 5 |
Workflow Clarity | A clear 3-step sequence with an explicit conditional fallback ("only if MCP tool call fails") and a file-existence guard; this is a non-destructive diagnostics skill so the validation cap does not apply, though there is no result-verification checkpoint. | 4 / 5 |
Progressive Disclosure | Under 50 lines with no external references needed, and the content is organized into a clear intro and an Instructions section, satisfying the simple-skill exception for progressive disclosure. | 5 / 5 |
Total | 17 / 20 Passed |