Content
60%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 a well-structured, governance-oriented instruction skill with a clear procedure, validation section, and real bundle references. Its weaknesses are redundant safety/scope sections and unspecified validation checks that limit actionability and conciseness.
Suggestions
Name the concrete structure checks in the Validation section (or point to the contract/evals files) and spell out a validate-fix-retry loop so the workflow is fully executable.
Consolidate the overlapping scope and safety guidance across Philosophy, Constraints, Anti-patterns, Failure mode, and Gotchas into fewer sections to reduce redundancy.
Explicitly signpost references/contract.yaml, evals.yaml, and task-profile.json from the body so every bundle file is reachable from the overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and teaches no background concepts, but Philosophy, Constraints, Anti-patterns, Failure mode, and Gotchas redundantly restate overlapping scope and safety guidance that could be consolidated. | 3 / 5 |
Actionability | Concrete file references and required sections are specified, but 'run the relevant structure checks' is unnamed and no executable verification step or command is given, leaving key details missing. | 3 / 5 |
Workflow Clarity | A clearly numbered 6-step procedure pairs with a dedicated Validation section and a fail-fast checkpoint; the gap is that the validate-fix-retry loop is implicit rather than spelled out and the checks are not named. | 4 / 5 |
Progressive Disclosure | The overview is lean with well-signaled one-level-deep 'Read when:' references to real files, though contract.yaml, evals.yaml, and task-profile.json are never explicitly signposted from the body. | 4 / 5 |
Total | 14 / 20 Passed |