Content
62%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-structured skill with clear workflow sequencing and good use of templates for verdicts and reports. Its main weaknesses are moderate verbosity (mermaid diagrams, some redundant framing) and incomplete actionability — the core mechanism for spawning parallel verification agents is described conceptually rather than with concrete, executable instructions. The evidence rules and CoVe protocol are valuable additions that demonstrate thoughtful design.
Suggestions
Replace mermaid diagrams with concise numbered lists or brief text descriptions — Claude processes text more efficiently than visual flowcharts
Add concrete instructions for how to spawn '@fact-checker' agents (e.g., the specific tool call or subagent mechanism to use), rather than describing it abstractly
Tighten the claim classification section by combining it with the verification agent prompt template, eliminating the redundant enumeration of the same fields
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is reasonably efficient but includes some content that could be tightened — the mermaid diagrams add visual bulk without much value for Claude (who processes text), and some sections like 'When NOT to Use' and the evidence rules preamble add moderate padding. The claim classification and wave execution sections could be more compact. | 2 / 3 |
Actionability | The skill provides structured templates (verdict format, agent prompt, report format) and specific post-action commands, which is good. However, the core verification process relies on spawning '@fact-checker' agents without explaining how to actually spawn them (what tool/mechanism), and the claim extraction process is described abstractly rather than with concrete executable steps. The CLI commands like 'backlog list --format json' and 'uv run prek run' are specific but their availability/context is assumed without verification. | 2 / 3 |
Workflow Clarity | The multi-step workflow is clearly sequenced: extract claims → classify → spawn agents in waves → collect verdicts → generate report → post-actions. The Chain of Verification (CoVe) requirement adds an explicit validation/feedback loop within each agent's process. The wave execution pattern handles different claim counts. Post-actions include linting and committing with specific commands. | 3 / 3 |
Progressive Disclosure | The skill references four external skills in the References section with relative paths, providing one-level-deep navigation. However, no bundle files are provided to verify these references exist, and the skill itself is fairly long (~150 lines of substantive content) with sections like the full report template and mermaid diagrams that could potentially be split out. The content is well-sectioned with clear headers, but it's borderline monolithic. | 2 / 3 |
Total | 9 / 12 Passed |