Content
60%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 presents a well-sequenced 10-step debugging workflow with explicit validation criteria, but is dragged down by circular filler sections, a persona statement, and a pseudocode example built on imaginary functions. Guidance names specific tools and techniques yet rarely gives executable commands, leaving it more descriptive than instructive.
Suggestions
Delete the circular 'Use this skill when'/'Do not use this skill when' sections and the generic 'Instructions' bullets that merely restate the skill name.
Replace the pseudocode example (aiAnalyze, getSentryIssue, getDataDogTraces) with executable commands or real queries, or convert it into concrete non-code instructions.
Drop the persona line ('You are an expert AI-assisted debugging specialist...') and the trailing 'Issue to debug: [user request]' template remnant to tighten token efficiency.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The workflow core is dense telegraphic lists, but several padded sections remain: "Working on error diagnostics smart debug tasks or workflows" (circular restatement of the skill name), "Apply relevant best practices and validate outcomes" (generic filler), the persona line "You are an expert AI-assisted debugging specialist", and the trailing "Issue to debug: [user request]" template remnant. Not 2 because there is little over-explanation of concepts Claude already knows; not 4 because the filler sections are clearly trimmable. | 3 / 5 |
Actionability | Specific tools ("Sentry, Rollbar, Bugsnag", "rr/Redux DevTools", "Pyroscope") and techniques ("Canary deploy debug version to 10% traffic") provide real guidance, but the example is pseudocode with non-existent functions (aiAnalyze, getSentryIssue, getDataDogTraces) and many steps describe rather than instruct ("AI suggests optimal breakpoint/logpoint locations"). Not 4 because no executable code or commands are given; not 2 because the tool and technique specifics are concretely actionable. | 3 / 5 |
Workflow Clarity | A 10-step numbered workflow with an explicit validation step and success criteria ("Tests pass", "No performance regression", "Error rate unchanged or decreased") plus falsification criteria for hypotheses. Not 5 because there is no explicit error-recovery loop (e.g., what to do when validation fails or a hypothesis is falsified mid-fix); not 3 because validation checkpoints are explicitly present, not merely implied. | 4 / 5 |
Progressive Disclosure | A single self-contained file with clear section headers (Triage → Observability → Hypotheses → ... → Prevention) and no buried or nested references. Not 5 because at ~194 lines the tool catalogs, production-safety techniques, and worked example could be split into reference files to keep SKILL.md lean; not 3 because the structure is good and content is mostly appropriately placed for a cohesive workflow. | 4 / 5 |
Total | 14 / 20 Passed |