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.
An exemplary workflow skill: steps are strictly ordered and gated, commands are copy-paste ready, fallbacks for missing capabilities are defined up front, and detail is delegated to a real one-level-deep reference file. The only trimmable fat is a few extended rationale passages, and a couple of rule lists (observability naming, issue body) could be extracted to reference files.
Suggestions
Move the observability naming and span/metric rules to a reference file (e.g. references/observability.md, or lean harder on wiki/observability/overview.md) and keep only the one-line requirement in the plan checklist — the current inline list is the densest block in the body.
Tighten the two longest rationale passages (the complexity-contamination explanation in Step 0 and the NEW-FEAT.md compaction paragraph) to their operative rules; the reasoning is persuasive but the commands and constraints are what the executor needs.
Consider extracting the issue-body template and label/type rules to a reference so Step 0 reads as compactly as Step 1.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body assumes Claude's competence throughout — no basic-concept explanations — and nearly every sentence carries project-specific constraints ('a wrong config value produces a tidy diff, a green build and no behaviour change, which no gate can fail on'). A few rationale passages run long (the complexity-rating justification, the NEW-FEAT.md blackboard paragraph) and could be trimmed, which is the 4 anchor's 'minor instances of over-explanation' rather than the uniformly lean 5. | 4 / 5 |
Actionability | Fully executable throughout: the pre-flight capability probe, 'gh issue create' with a complete body heredoc, both the worktree and in-place branch recipes with exact commands including the 'git config branch.<branch>.gh-merge-base' trick for stacked PRs — copy-paste ready with placeholders only where user input is genuinely required, matching the 5 anchor. | 5 / 5 |
Workflow Clarity | 'Four things, in this order — each depends on the one before' plus Steps 0–4 gives an explicit sequence; validation is equally explicit — the capability probe runs 'once, before Step 0' rather than failing per step, the 'When a gate cannot run' section defines fallback and honesty rules, and stress-plan is a hard gate ('every finding closed, before a line of code is written'). This is the 5 anchor's clear sequence with explicit validation and error-recovery, not 4's 'minor validation gaps'. | 5 / 5 |
Progressive Disclosure | The body is a well-organized overview that delegates cleanly: 'The routing table is references/skills-routing.md' points to a real one-level-deep bundle file, and detailed material (rate-complexity, stress-plan, prep-pr, wiki/observability/overview.md) is referenced rather than inlined. Minor gaps keep it at 4 — the observability naming rules and issue-body conventions are long enough to live in a reference file, and skills-routing.md is the only bundle file carrying that load. | 4 / 5 |
Total | 18 / 20 Passed |