Content
70%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The content is well-organized with clear sections, concrete OpenSpec anchors, and an unambiguous stance-plus-conditional-flow structure. Its weaknesses are conciseness (padded diagrams and illustrative dialogs) and actionability (a fair amount of abstract behavioral coaching alongside the concrete guidance).
Suggestions
Trim the decorative ASCII boxes and reduce the four full dialog walkthroughs to one or two compact examples to reclaim token budget.
Pair each abstract behavioral directive (e.g. 'ask questions that emerge naturally') with a concrete anchor or micro-example so the guidance is actionable rather than coaching.
Consider moving the multi-entry-point dialog examples into a reference file to keep SKILL.md a lean overview, since the file is ~280 lines.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient and avoids lecturing on basics Claude already knows, but it is padded with large decorative ASCII boxes and four full illustrative dialog walkthroughs that could be tightened without losing clarity. | 2 / 3 |
Actionability | Concrete anchors exist (openspec list --json, artifact paths, the insight-to-artifact capture table, slash commands), but much of the guidance is abstract behavioral coaching ('ask questions that emerge naturally') rather than concrete steps. | 2 / 3 |
Workflow Clarity | It deliberately declares itself a non-workflow, and the one structured sequence ('When a change exists') is clearly numbered with an explicit user-decides checkpoint; no validation gaps apply since destructive/batch operations are explicitly forbidden. | 3 / 3 |
Progressive Disclosure | It is a single self-contained file with no bundle files or nested references, organized into cleanly headed sections with horizontal rules, giving easy navigation. | 3 / 3 |
Total | 10 / 12 Passed |