Content
82%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 body is a well-organized, highly actionable command reference with real verification steps in its worked examples and genuine one-level-deep references. Its main weaknesses are moderate duplication between the wrapper-script and direct-glab command listings and the absence of pre-flight checks for destructive operations.
Suggestions
Trim the 'Commands (Direct glab Usage)' catalog: keep the 2-3 most common commands per section inline and move the full listing to a reference file (as done with common-workflows.md), since each section already links to the official glab docs.
Add explicit pre-flight validation before destructive actions — e.g., check 'glab ci status' and MR state before 'glab mr merge 456', and verify outcome after merge/close/delete — to close the feedback loop the bare catalog commands lack.
Consolidate the duplicated list/view operations: a small table mapping read operations to the wrapper script and write operations to direct glab (complementing references/permissions.md) would replace two parallel command listings and cut token cost.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient — terse command listings with inline comments, no conceptual padding about GitLab or CI/CD basics. Not 5: the ~120-line direct-glab command catalog (issues, mrs, ci, repos, releases, config, aliases) substantially duplicates operations already shown via the wrapper script, and each section also links to official docs, so some listings could be trimmed. | 4 / 5 |
Actionability | Commands are copy-paste ready with realistic flags ('--assignee=@me', '--title "Fix login bug" --description "Closes #123"'), and the Examples section walks through concrete multi-step cases (daily MR review, issue + linked MR, pipeline monitoring). Common cases are covered with executable commands throughout. | 5 / 5 |
Workflow Clarity | Example workflows include explicit verification checkpoints ('# Verify: note the issue number from output', '# Verify approval was recorded', '# Verify pipeline restarted'). Not 5: destructive commands ('glab mr merge 456', 'glab issue close 123', 'glab release delete v1.0.0') appear as bare catalog entries without pre-flight checks (e.g., 'glab ci status' before merge) or error-recovery guidance. | 4 / 5 |
Progressive Disclosure | Structure is good: both referenced files (references/permissions.md, references/common-workflows.md) exist, are one level deep, and are clearly signaled with markdown links; the read-vs-write split between wrapper script and direct glab is explained and documented in permissions.md. Not 5: the full direct-glab command catalog is inlined in SKILL.md where a reference file (like the existing common-workflows.md pattern) would keep the overview leaner. | 4 / 5 |
Total | 17 / 20 Passed |