Content
88%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.
A well-engineered operational skill: an explicitly sequenced six-step workflow with genuine validation checkpoints and feedback loops around its destructive paths, fully executable commands, and exact interaction scripts. It is only held back from perfection by light redundancy between the steps and the rationalizations table, and by keeping everything — including supplementary material — inline in one file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with commands, exact menu text, and decision tables rather than concept explanations, but has minor trimmable redundancy: the MAIN_ROOT computation and cd appear twice (Option 1 and the discard path), and the 'Common Rationalizations' table restates safety rules already given in Steps 4–6. This matches 'efficient; minor instances of over-explanation that could be trimmed' rather than the every-token-earns-its-place anchor 5. | 4 / 5 |
Actionability | Nearly everything is copy-paste-ready bash (environment detection, merge sequence, worktree removal, status --porcelain -uall), exact literal menu prompts, and an exact typed-'discard' confirmation protocol; placeholders (<feature-branch>, <base-branch>) are appropriate, and the one flexible step (PR creation 'with the forge's tooling — its CLI if one is available, or the creation URL most forges print') explicitly justifies its flexibility across forges. | 5 / 5 |
Workflow Clarity | The core principle line states the sequence (Verify tests → Detect environment → Present options → Execute choice → Clean up), each step is numbered with explicit validation: tests must pass before the menu, the merged result is re-tested with a stop-and-investigate path, worktree removal refusal triggers a show-files-and-ask feedback loop, and deletion requires a typed 'discard'. This is exactly the clear-sequence-with-explicit-validation-and-feedback-loops anchor, which matters for a skill with destructive operations. | 5 / 5 |
Progressive Disclosure | No bundle files exist and the body references none, so all content is appropriately self-contained with well-organized sections, tables, and a Quick Reference; however, at ~220 lines it is a single-file monolith whose 'Common Rationalizations' table (10 rows) is the kind of supplementary material that could live in a one-level-deep reference file, leaving minor organization gaps versus the clean split/navigation of anchor 5. | 4 / 5 |
Total | 18 / 20 Passed |