Content
96%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.
An excellent single-file skill: dense with non-obvious domain rules, fully executable query recipes covering the common cases, and explicit guidance on interpreting partial or permission-limited results. The only structural improvement would be splitting the recipes into a reference file to keep SKILL.md leaner.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every rule conveys non-obvious domain knowledge (status filter values, 'STARTS WITH' vs. fixed-depth group walking, SBOM match semantics, Security Manager role effects, pagination traps) with no padding and no re-explanation of concepts Claude already knows. This matches the anchor 'every token earns its place'. | 5 / 5 |
Actionability | Nine complete, executable 'gql orbit-query' blocks with concrete real-world names and versions, plus exact 'glab api' shell commands for tag counts. The guidance is copy-paste ready and the recipes cover the common question types end to end. | 5 / 5 |
Workflow Clarity | As a single-purpose, read-only query skill the action is unambiguous ('Run each recipe with glab orbit query'), each task type has a named recipe, and result validation is addressed ('a response with has_more: true or truncated: true... Group and count instead, or page with SKIP') along with safety caveats about missing vulnerability links. No destructive or batch operations, so no validation cap applies. | 5 / 5 |
Progressive Disclosure | The body is well-organized (Rules vs. Recipes) with working internal anchor links and cohesive placement, but at roughly 130 lines with everything inline it exceeds the under-50-line threshold that would let a sectioned single file score 5. It is not anchor 3 because the structure is clear and navigable, and no content is buried. | 4 / 5 |
Total | 19 / 20 Passed |