Content
85%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-sequenced, highly actionable workflow with strong validation checkpoints, though slightly redundant in its hard-constraint boilerplate. Its main weakness is reliance on a comet/reference tree that is not bundled here, leaving progressive disclosure partially dangling.
Suggestions
Ship the referenced comet/reference/*.md files (decision-point, subagent-dispatch, debug-gate, context-recovery, dirty-worktree, auto-transition) as actual bundle files so the references resolve.
Consolidate the repeated 'script-enforced hard constraint' explanations for isolation/subagent_dispatch/tdd_mode/build_mode into one table to reduce boilerplate.
Add a one-line note near the top listing the reference files this skill depends on, so the dependency surface is visible before the reader encounters each inline reference.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is largely efficient instructional prose that assumes competence and delegates detail to reference files, with only minor repeated hard-constraint boilerplate that could be trimmed. | 4 / 5 |
Actionability | It provides concrete, copy-paste-ready bash commands (comet-state, git checkout -b, git rev-parse HEAD) and explicit skill-load instructions throughout, fully executable. | 5 / 5 |
Workflow Clarity | Steps 0–5 are clearly sequenced with explicit validation checkpoints (entry check, review gate, debug gate) and feedback loops (fix and re-validate) for fragile operations. | 5 / 5 |
Progressive Disclosure | It signals one-level-deep references to comet/reference/*.md files, but none of those bundle files actually exist in the skill directory, so the references are present but unverifiable and key detail is left inline. | 3 / 5 |
Total | 17 / 20 Passed |