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.
A highly actionable, well-sequenced workflow with concrete commands, exact JSON fields, and explicit validation and stop-condition checkpoints for a destructive merge operation. The only real cost is length: duplicated guardrails and deep edge-case digressions (retirement conditions, store selection) make it longer than the task strictly requires.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, operational edge-case guidance that assumes Claude's competence (no concept padding), but the Guardrails section restates rules already given in steps 3-4 ("Use only artifactPaths.specs.existingOutputPaths; never infer...", "Honor a caller-supplied subset... never widen it back to the full list") and "Never write an empty ## Requirements section" appears twice, so minor trimming is possible. | 4 / 5 |
Actionability | Provides fully executable commands (openspec status --change "<name>" --json, openspec list --json, openspec instructions specs --change "<name>" --json, openspec validate --specs), exact JSON fields to read (planningHome.root, artifactPaths.specs.existingOutputPaths), exact file paths, copy-paste format templates, and a concrete success-output template covering the common cases. | 5 / 5 |
Workflow Clarity | Six clearly sequenced steps with an explicit validation checkpoint (step 5: run openspec validate --specs and do not claim success on failure) and extensive pre-write gates (root check, "Declared in" error branch, stop on non-zero/invalid instructions JSON, six retirement conditions), giving every risky write an explicit check. | 5 / 5 |
Progressive Disclosure | A single-file skill (no references/, scripts/, or assets/ exist) with well-organized, clearly headed sections (Steps, Delta/Main Spec Format References, Key Principle, Output, Guardrails); the format references are used mid-workflow so inline placement is defensible, but at ~280 lines the store-selection and retirement digressions could live in referenced files. | 4 / 5 |
Total | 18 / 20 Passed |