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 strong, highly actionable single-file workflow: exact CLI commands, parse markers, issue templates, and a degradation ladder make it essentially copy-paste executable with clear validation checkpoints. The only real weakness is redundancy — the "Issues by Priority" and parts of the "Verification Heuristics" sections restate what the numbered steps already prescribe.
Suggestions
Collapse the "Issues by Priority" section into a one-line severity definition (CRITICAL = must fix before archive, WARNING = should fix, SUGGESTION = nice to fix) and drop the re-enumeration of issue contents already given in steps 5–7 — this trims ~15 redundant lines.
Trim the "Verification Heuristics" lines that restate step guidance (e.g., 'Completeness: Focus on objective checklist items') and keep only the genuinely new rules (false-positive preference, actionability requirement).
Consider moving the verification report template (the scorecard table and final-assessment wording) into a short reference or asset file so SKILL.md stays a lean process overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient procedural instruction with no explanation of concepts Claude already knows, but it can be noticeably tightened: the "Issues by Priority" section re-enumerates the CRITICAL/WARNING/SUGGESTION contents already fully specified in steps 5–7 ('Incomplete tasks', 'Missing requirement implementations', 'Spec/design divergences', 'Missing scenario coverage'), and parts of "Verification Heuristics" ('Focus on objective checklist items', 'Use keyword search...') restate the step instructions. This is anchor 3 ('could be tightened') rather than anchor 4's 'minor instances' — the redundancy spans whole sections, not isolated lines — and well above anchor 2's padded explanations. | 3 / 5 |
Actionability | Guidance is fully executable: exact copy-paste commands ('openspec list --json', 'openspec status --change "<name>" --json', 'openspec instructions apply --change "<name>" --json'), exact artifact paths ('openspec/changes/<name>/specs/'), exact parse markers ('### Requirement:', '#### Scenario:', '- [ ]' vs '- [x]'), a fixed issue taxonomy with literal message templates ('Requirement not found: <requirement name>'), and a concrete output format. Per the rubric's instruction-only-skill note, the absence of code is not penalized since the guidance is copy-paste-level actionable, matching anchor 5. | 5 / 5 |
Workflow Clarity | Eight numbered steps in a clear sequence with explicit input handling up front ('Do NOT guess or auto-select a change'), a graceful-degradation ladder for missing artifacts ('If only tasks.md exists: verify task completion only... Always note which checks were skipped'), and an explicit final-assessment decision rule ('If CRITICAL issues... If only warnings... If all clear'). These are the validation checkpoints, error-recovery paths, and checklists anchor 5 requires; the skill is itself a verification workflow, so no validation gap remains, keeping it above anchor 4. | 5 / 5 |
Progressive Disclosure | No bundle files exist and the body is well organized with numbered steps and distinct sections (Verification Heuristics, Graceful Degradation, Output Format), so nothing is buried. However, at ~155 lines it exceeds the simple-skill threshold that would let a single well-organized file score 5, and content like the report template and the per-dimension issue-generation rules could arguably live in reference files. That leaves minor organization/split opportunities, fitting anchor 4 ('Good structure; most content is appropriately placed; minor organization gaps') rather than anchor 5's 'appropriately split' ideal. | 4 / 5 |
Total | 17 / 20 Passed |