Content
22%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is excessively verbose, containing extensive sample outputs, CI/CD configurations, custom analysis scripts, and categorical lists that inflate the document without proportional value. While the CLI commands are concrete and actionable, the skill lacks proper multi-step workflows with validation checkpoints—critical for a performance analysis tool where misdiagnosis could lead to counterproductive optimizations. The content would benefit greatly from aggressive trimming and restructuring into a concise overview with referenced detail files.
Suggestions
Cut the document to under 100 lines by removing sample outputs, CI/CD YAML, custom scripts, and metric category lists—move these to referenced bundle files like EXAMPLES.md and CI-INTEGRATION.md
Add explicit sequenced workflows with validation steps, e.g.: 1. Run detection → 2. Review output → 3. Validate recommendations against current config → 4. Apply fixes incrementally → 5. Re-run detection to confirm improvement
Remove descriptive content Claude already knows (what cache hit rates are, what memory bottlenecks mean) and replace with decision-making heuristics (e.g., 'If utilization < 60%, reduce agent count before investigating other bottlenecks')
Consolidate the redundant output examples (CLI text output, JSON format, markdown report) into a single canonical example with a reference to a formats file
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~400+ lines. Contains extensive lists of metrics categories, sample outputs, CI/CD YAML, custom scripts, best practices, and troubleshooting sections that are largely padded content. Much of this (e.g., explaining what 'cache hit rates' or 'memory access patterns' are, listing every possible bottleneck category) is knowledge Claude already has. The sample report output, sample JSON, and sample markdown report are all redundant ways of showing the same information. | 1 / 3 |
Actionability | The CLI commands are concrete and copy-paste ready (e.g., `npx claude-flow bottleneck detect --fix --threshold 15`), and the MCP integration shows specific function calls. However, much of the content is descriptive rather than instructive—listing what metrics are analyzed, what report sections exist, and what typical improvements look like, without giving Claude clear decision-making guidance on when to use which approach. | 2 / 3 |
Workflow Clarity | Despite being a complex multi-step analysis skill, there are no clear sequenced workflows with validation checkpoints. The 'Analyze and Auto-Fix' is a single command, not a workflow. The best practices section says 'always review before applying --fix' but never shows a concrete review-then-fix workflow. The troubleshooting section lists commands without sequencing them into diagnostic workflows with decision points. | 1 / 3 |
Progressive Disclosure | The document references several external files in 'See Also' (bottleneck-detect.md, performance-report.md, etc.) and cross-references other skills, which is good structure. However, no bundle files are provided to verify these references exist, and the main SKILL.md itself is monolithic—containing extensive sample outputs, CI/CD configs, and custom scripts that should be in separate referenced files rather than inline. | 2 / 3 |
Total | 6 / 12 Passed |