Content
81%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.
A highly actionable, well-sequenced skill body with exact commands, a validation ladder, and a properly split reference layer. Its weaknesses are redundancy (First-Principles Gate boilerplate, repeated blocked-plan rules) and references that assume an external repo layout.
Suggestions
Compress or remove the First-Principles Gate boilerplate (desired outcome / rejected assumptions / evidence required bullets) that restates constraints already covered in Constraints, Workflow, and Failure Mode.
Consolidate the repeated read-only/blocked-plan guidance (Workflow step 9-10 paragraph, Execution Boundaries, Failure Mode) into a single section to reclaim tokens.
Clarify or localize the external repo-root references ('Infrastructure/references/...', 'Plugins/skill-factory/...'), which do not resolve from the skill package and conflict with the skill's own portability constraint.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly telegraphic and free of concept-teaching padding, but the ~11-bullet First-Principles Gate reads as compliance boilerplate rather than actionable guidance, and the read-only/blocked-plan rules are repeated across Workflow, Execution Boundaries, and Failure Mode. Mostly efficient but could be tightened fits anchor 3; it is not anchor 2 because there is no explanation of things Claude already knows, and not anchor 4 because the redundancy is more than minor. | 3 / 5 |
Actionability | Guidance is fully executable: twelve exact './bin/ask ... --json --robot' validation commands, the exact minimal package file list ('SKILL.md', 'references/contract.yaml', 'references/evals.yaml', 'references/task-profile.json'), and a copy-paste YAML output template with concrete fields. Copy-paste ready commands covering the common cases match anchor 5. | 5 / 5 |
Workflow Clarity | The 10-step Workflow is clearly sequenced with explicit validation checkpoints ('Run strict audit. If it fails, fix that failure before broader validation'), and the Validation section defines an ordered proof ladder with a fail-fast rule ('stop at the first failed gate'). Clear sequence with explicit validation steps and error-recovery loops matches anchor 5. | 5 / 5 |
Progressive Disclosure | SKILL.md is a genuine overview and the bulk content (contract.yaml, evals.yaml, task-profile.json, template, examples, interview) is split one level deep into well-signaled references that all exist in the bundle. However, several References-section entries point outside the bundle to repo-root paths ('Infrastructure/references/first-principles-factory-gate.md', 'Plugins/skill-factory/references/context-development-lifecycle.md', lens packs) that do not resolve from the skill package itself — a minor organization gap consistent with anchor 4 rather than anchor 5. | 4 / 5 |
Total | 17 / 20 Passed |