Content
100%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.
An exemplary instruction-only skill body: dense and project-specific with zero filler, fully executable commands and templates, a checklist-plus-procedure structure with pre-flight states and explicit termination, and clean delegation of the checklist detail to CHECKLIST.md. Nothing scored below the top anchor on any dimension.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence throughout — it cites Ousterhout's red flags, ADR-0008, and shared invariants without explaining them, and every rationale line is kit-specific (e.g. "UTC, because the advisory that reads the row compares against UTC midnight") rather than general knowledge, so no token is spent on what Claude already knows. | 5 / 5 |
Actionability | Guidance is copy-paste ready: `sh scripts/agents.lib.sh planner`, the full trace-emit command with its fallback `|| :`, the diary row template `| **Last housekeeping** | <today, ISO, UTC> — …`, and the report path format `<tmpdir>/housekeeping-<YYYYMMDDTHHMMSSZ>.md`, with concrete routing per finding class. | 5 / 5 |
Workflow Clarity | The six-step procedure is clearly sequenced with explicit checkpoints: three pre-flight states (first pass, unstamped article, nothing to prune), per-item evidence requirements ("A pass that finds nothing says so, per item"), a defined stop condition, and an anti-patterns section covering failure modes; the pass is read-only, so no destructive-operation cap applies. | 5 / 5 |
Progressive Disclosure | No bundle files were provided to verify, but scored on the body's structure: it is a well-organized overview that delegates the full checklist form to CHECKLIST.md with a clear signal ("The full form of each — what to look at, what counts as a finding, what to route it to — is in CHECKLIST.md"), a single well-signaled, one-level-deep reference. | 5 / 5 |
Total | 20 / 20 Passed |