Content
77%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 disciplined, token-efficient gate protocol with excellent sequencing and per-phase validation checkpoints that fully cover the batch-merge risk profile. Its main weaknesses are execution-level detail (few runnable commands; most steps are stated as outcomes rather than operations) and a progressive-disclosure design that leans on shared files that are not present in the skill's bundle.
Suggestions
Add the concrete commands for the recurring operations (e.g. the gh/git commands to 'resolve each selected PR's open/draft state, current head/base' and the exact 'applicable static profile' invocations), or point to the specific shared file section that defines them.
Ship or verify the referenced files inside the skill bundle: `../_shared/base-green.md`, `../_shared/validation-gate.md`, `../_shared/merge-gates.md`, and `scripts/quality/check-skill-contract.mjs` currently do not exist relative to this SKILL.md, so the delegated protocol cannot actually be loaded.
Include a brief filled-in example of the '1–5 star table with rationale, overlaps and holds' so the intake artifact has a concrete shape rather than only a description.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and imperative throughout — e.g. "Pin the approved live base and record the ordered PR heads", "Keep stdout/stderr, exits, signals, versions and identities", "never use git stash" — with zero padding and no explanation of concepts Claude already knows. Every line states a rule or constraint, matching the 'every token earns its place' anchor; it is not the level-4 case because there are no over-explanations to trim. | 5 / 5 |
Actionability | Some concrete guidance exists — "Run `scripts/quality/check-skill-contract.mjs`", "one owned branch and worktree under `.claude/worktrees/`", "`gh pr merge --match-head-commit` guards the head only" — but most steps stop short of executable detail ("Resolve each selected PR's open/draft state" and "Run the applicable static profile" name no commands), so key execution details are missing. It is above level 2 (it is more than high-level hints: specific paths, flags, and artifacts are given) but below level 4 because the majority of operations lack concrete commands or examples. | 3 / 5 |
Workflow Clarity | Five explicitly sequenced phases each end with a validation checkpoint ("Completion: selected scope, target, ownership and permitted actions are explicit") and error-recovery feedback loops are pervasive: "A changed identity returns the item to validation", "a later PASS does not erase an earlier required FAIL", and section 5's re-validation of the resulting base before the next landing. This satisfies the batch-operation validation requirement and matches the anchor with explicit validation steps, feedback loops, and checklists; no minor gaps are evident. | 5 / 5 |
Progressive Disclosure | The body opens by delegating to clearly signaled references ("Read `../_shared/base-green.md`, `../_shared/validation-gate.md` and `../_shared/merge-gates.md`") and sections are well organized, but none of the referenced files exist in this skill's bundle — there is no `references/`, `scripts/`, or `assets/` directory, and `scripts/quality/check-skill-contract.mjs` plus the `../_shared/` files resolve outside the skill and cannot be verified. This falls short of level 4 ('minor organization gaps') because the skill's core protocol depends on dangling, unverifiable paths, yet stays above level 2 because references are explicit and the inline content is structured. | 3 / 5 |
Total | 16 / 20 Passed |