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 strong, dense CLI skill body: fully executable command guidance, clearly sequenced workflows with genuine validation and error-recovery loops, and no wasted explanation of concepts Claude already knows. The main improvement opportunities are consolidating repeated guidance and moving some inline detail into the existing reference files with explicit links.
Suggestions
Consolidate the `--status-after` guidance, which is repeated in ~7 places, into the IDs/Chaining section and reference it once from the recipes.
Link references/concepts.md and references/examples.md directly (e.g., "See [concepts.md](references/concepts.md)") alongside the `but skill concepts` / `but skill examples` commands so the bundle files are discoverable from the body.
Consider moving the long linked-worktree and resolve-conflicts recipe details into a reference file, keeping a condensed core loop in SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Assumes Claude's git competence throughout — every rule conveys tool-specific semantics (ID lifetimes, refusal behavior, chaining rules) with no padding or basic-concept explanations. Not 5: the `--status-after` guidance is repeated across roughly seven sections and "but pr new... do not run but push before it" appears twice, so a few instances could be consolidated; not 3: there is no unnecessary explanation beyond that deliberate reinforcement. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready commands with exact syntax throughout: "but commit -b <branch> -m \"<msg>\" <id> <id>", concrete ID examples ("uvw:2e4", "mzm:uvw:2e4"), quoted error strings ("Unclear where to commit. Found more than one stack"), and a fast path for the common selective-commit case. Specific examples cover the common cases comprehensively. | 5 / 5 |
Workflow Clarity | Multi-step recipes are clearly sequenced with explicit validation and feedback loops: the conflict loop ("Go back to step 1 until it reports no conflicts"), `but pull --check` as a preview, `but resolve finish` reporting leftover markers and remaining conflicts, recovery procedures for failed pushes and dependency refusals, and inverse commands for destructive operations. Error recovery is addressed at every fragile step. | 5 / 5 |
Progressive Disclosure | Bundle references (references/concepts.md, references/examples.md) are real, one level deep, and clearly signaled ("For workspace model: `but skill concepts`", "For workflow examples: `but skill examples`"). Not 5: pointers are CLI commands rather than file links, the "but skill reference" pointer has no corresponding bundle file, and the ~250-line body carries dense operational detail inline rather than splitting it into the reference files. | 4 / 5 |
Total | 18 / 20 Passed |