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.
This is a well-engineered destructive-operation skill: lean, fully actionable, with thorough validation checkpoints and feedback loops that protect against data loss. The only marginal gaps are minor conciseness trims and the deliberately forge-agnostic PR step.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean, assumes Claude's competence (no explanations of git/worktrees), and gives purposeful code comments; only minor trim opportunities exist such as the 'Announce at start' line and the length of the Common Rationalizations table. | 4 / 5 |
Actionability | Concrete, executable bash is provided throughout (GIT_DIR capture, git push, worktree remove) with clearly templated placeholders; the only soft spot is the forge-agnostic PR-creation step, which is intentionally left flexible rather than copy-paste. | 4 / 5 |
Workflow Clarity | A clearly sequenced six-step workflow is packed with explicit validation checkpoints (stop on red tests, verify merged result, typed 'discard' confirmation, worktree-removal refusal handling) plus a Quick Reference table and Common Rationalizations that function as checklists — satisfying the destructive-skill validation requirement. | 5 / 5 |
Progressive Disclosure | No bundle files exist and the skill is a cohesive self-contained workflow; the single SKILL.md is well-organized with clear Step 1–6 headers plus reference tables, which is appropriate structure for a skill with no need for external references. | 5 / 5 |
Total | 18 / 20 Passed |