Content
71%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 unusually dense, high-signal process skill: nearly every sentence states a specific rule, check, or template, and validation checkpoints (`br ready --json`, the audit script, twice-verified structural claims) are woven throughout. The main weaknesses are a dangling script reference with no bundle behind it, inlined template content that belongs in a reference file, and anecdotal padding that could be tightened.
Suggestions
Ship the referenced scripts/tracker-audit.bb in a scripts/ directory (or drop the path from the Mechanical audit section) — the body instructs running a file that does not exist in the bundle.
Move the eight-section bead anatomy template (plus the Risk/DO NOT extensions) into a references/ file and keep a one-line summary in SKILL.md, shortening the main file and making the template directly linkable from other beads.
Trim the field-evidence anecdotes (the 92-bead port, the "slice 1.5" relocation) to a single clause each; the rules they justify are already self-contained.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense, rule-per-sentence prose that assumes an intelligent reader ("the anti-NIH payload", "Negative results are the most expensive knowledge to regenerate"). Minor trimmable padding: the field-evidence anecdotes ("a 92-bead port closed 74 of them in four days", the "slice 1.5" relocation story) justify rules but run long. | 4 / 5 |
Actionability | Gives concrete, executable guidance: specific checks ("retarget the dep to the specific sub-bead, verify with `br ready --json`"), a command (`scripts/tracker-audit.bb [issues.jsonl]`), and filled-in templates (FOUNDATION WARNING block, close-reason contents). Minor gap: the referenced audit script does not exist in the bundle, and quoted examples are illustrative rather than copy-paste ready. | 4 / 5 |
Workflow Clarity | Each core procedure is clearly sequenced with checkpoints: the deferral rule (stop → search the tracker → create the bead → reference its id in the close-reason), dependency audit with `br ready --json` verification, and the mechanical audit with documented false-positive triage. Falls short of 5 because the skill is a set of parallel disciplines rather than one end-to-end workflow, and some checks (e.g., the twice-verified structural claim) describe the check without a command. | 4 / 5 |
Progressive Disclosure | Well-sectioned with clear headers, but it is a ~175-line single file with the full bead-anatomy template inlined that could live in a reference file, and it cites "scripts/tracker-audit.bb" while no scripts/ directory exists in the bundle — a dangling reference. This matches the anchor of having structure with content that should be separate left inline. | 3 / 5 |
Total | 15 / 20 Passed |