Content
88%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 content is highly actionable with executable commands, clearly sequenced workflows that include validation and recovery for batch operations, and well-organized sections. The main weakness is length — some repeated prose rules could be tightened, and the command reference could be split into a referenced file for cleaner progressive disclosure.
Suggestions
Tighten the repeated 'Do not/Never' rule sentences into a single consolidated constraints list to reduce token overhead.
Consider moving the full Command Reference and Review-controls flag tables into a referenced file (e.g., REFERENCES.md) with a concise summary inline, improving progressive disclosure.
The 'Slate V2 Operating Rules' section is project-specific; gating it behind a clear conditional trigger would keep the general skill leaner.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational — concrete commands and gotchas rather than concepts Claude already knows — but ~470 lines with repeated 'Do not/Never' prose rules could be trimmed in places, so it sits at efficient-with-minor-over-explanation rather than fully lean. | 4 / 5 |
Actionability | Provides copy-paste-ready commands throughout (e.g., 'clawpatch review --json', 'clawpatch fix --finding <findingId> --json') covering the common cases with specific flags and JSON output. | 5 / 5 |
Workflow Clarity | Multi-step processes are explicitly sequenced with validation checkpoints and feedback loops — the recoverable forced re-review JSONL loop and the 'Verification Closeout' (typecheck, lint, test, status) plus 'revalidate' satisfy the destructive/batch validation requirement. | 5 / 5 |
Progressive Disclosure | No bundle files exist and all content is inline, but the body is well-organized into clearly headed sections (Review, Fix, Revalidate, Locks, etc.) making it navigable; the long command reference could arguably split into a separate file, hence 4 not 5. | 4 / 5 |
Total | 18 / 20 Passed |