Content
81%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 strongly operational skill body: every step is backed by executable gh/git commands, the batch-merge workflow is sequenced with unusually thorough validation and failure branches, and external authority is clearly delegated. Minor deductions for trimmable rationale asides and a sample merge command that hardcodes a method the skill defers to another document.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational with no explanation of general concepts Claude already knows, but several rationale asides could be trimmed ("a local rebase would strip signatures, which is the usual reason a repo forbids rebase merges", "an untagged bump is a release no consumer can reach… the kit's own does, in its self-host suite's F3"). Not 5: these instances of over-explanation keep it short of lean; not 3: most of the rationale is repo-specific policy Claude could not know, not padding. | 4 / 5 |
Actionability | Nearly fully executable guidance: real, copy-paste-ready commands for every phase (`gh pr list --state open --json …`, `gh api -X PUT "repos/{owner}/{repo}/pulls/$PR/update-branch"`, `gh pr checks "$PR" --watch`, `gh run watch <databaseId>`). Not 5: the sample `gh pr merge "$PR" --merge` hardcodes a method the skill itself says is mandated elsewhere (`constitution/local-workflow.md`), a minor copy-paste hazard. | 4 / 5 |
Workflow Clarity | Clear five-step sequence (assemble → order → present plan → land each with sub-steps a–d → after batch) with extensive validation checkpoints: green-check verification, BEHIND detection, `--watch` on re-runs, explicit failure branches for 4b/merge-rejected/4d, and hard rule 6 (failed post-merge workflow stops the train). The hard-rules checklist and skip-to-/pr-iterate feedback loop match the top anchor for a batch operation. | 5 / 5 |
Progressive Disclosure | No bundle files exist; the skill is a single well-sectioned file with a dedicated Cross-references section pointing one level deep to real repo artifacts (`constitution/local-workflow.md`, `docs/adr/INDEX.md`) with what each provides. Not 5: the body is ~150 lines and monolithic — the ordering rationale and merge-method policy could arguably live in separate reference files; not 3: structure and signaling are clear, not buried. | 4 / 5 |
Total | 17 / 20 Passed |