Content
57%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 delivers concrete, runnable commands for all three tools and real-world use cases, but it is undermined by duplicated content, a feature-menu section that adds no instruction, and workflows that lack any validation or feedback checkpoints. With no reference files, everything from tool comparison to troubleshooting is inlined, making the skill heavier than it needs to be.
Suggestions
Add validation checkpoints to the workflows, e.g. after setup: 'Run semgrep --config=auto on the repo and confirm findings render as SARIF before wiring into CI', and in Next Steps: 'Verify the quality gate blocks on critical findings with a test scan'.
Remove the duplicated Quick Start code block under 'New Project Setup' (keep one canonical copy) and drop or compress the 'Overview'/'Core Capabilities' sections that restate the description.
Split per-tool depth (Semgrep rule authoring, SonarQube quality gate config, CodeQL query development) into references/ files linked from a lean overview, and turn vague Troubleshooting bullets into specific commands.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 'Overview' section repeats the frontmatter description almost verbatim, the Quick Start Semgrep commands ('pip install semgrep', 'semgrep --config=auto --error') are duplicated verbatim under 'New Project Setup', and 'Core Capabilities' is a descriptive feature menu rather than instruction. Not 2 because it never explains concepts Claude already knows and the code blocks are tight; not 4 because the duplication and padded bullet lists exceed minor trimming. | 3 / 5 |
Actionability | Mostly executable guidance with concrete, copy-paste commands ('docker run -d --name sonarqube -p 9000:9000 sonarqube:10.8-community', 'gh extension install github/gh-codeql', 'codeql database create mydb --language=python', 'semgrep --config p/pci-dss --json -o pci-scan-results.json') plus a complete custom-rule YAML. Not 5 because Troubleshooting and Best Practices give high-level direction ('Review and tune rule sensitivity', 'Parallelize scans across modules') without specific commands; not 3 because the majority of guidance is directly runnable. | 4 / 5 |
Workflow Clarity | Sequences are present and ordered ('Initial Assessment' steps 1-4, 'Next Steps' 1-6) but contain no validation checkpoints — nothing like verifying scan output, confirming a quality gate blocks on critical findings, or a fix-and-rerun loop. Not 2 because steps are clearly numbered and coherent; not 4 because checkpoints are wholly absent rather than having minor gaps. | 3 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent), so ~190 lines covering three tools, CI integration, compliance, and troubleshooting all sit inline in SKILL.md. Sections are well-labeled, but per-tool depth belongs in separate reference files and the only file links present are unverifiable cross-skill links. Not 4 because at this breadth the content should be split; not 2 because the document is structured and navigable. | 3 / 5 |
Total | 13 / 20 Passed |