Content
73%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-structured instruction-only skill with an unusually strong workflow: mandatory human gate before batch publishing, a trust boundary on untrusted PRD input, and copy-usable stamp formats. Its weakest dimension is token efficiency — the model-name prohibition and several rules are repeated across sections, and the anti-patterns list mostly restates earlier rules.
Suggestions
State the never-name-a-model rule once (rule 9 or the tier section) and drop the repetitions in the domain rubric and anti-patterns.
Trim the anti-patterns to only the entries not already covered by rules 1-10 (e.g., keep 'tickets that only make sense read together'; drop restatements of rules 4, 9, and 10).
Add one short example ticket body showing the title, 'Part of #<prd>', 'Blocked by:', 'Tier:', optional 'Domain:', and label assembled together.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes competence (no basic-concept explanations), but it could be tightened: the never-name-a-model rule is stated four times (rule 9, the tier section's 'Never write a model name', the domain rubric's 'Still never a model name', and the anti-patterns), and the anti-patterns bullets largely restate rules 4, 9, and 10. | 3 / 5 |
Actionability | Guidance is mostly executable — exact labels ('ready-for-agent'), exact stamp lines ('Blocked by: #N', 'Tier: <planner|implementer|mechanical|reviewer>', 'Domain: <token>' with a token regex), the concrete command 'gh issue create', and a first-hit-wins decision order for tiers. The minor gap is that no assembled example ticket body shows all the stamped lines together. | 4 / 5 |
Workflow Clarity | The Procedure is a clear 5-step sequence with an explicit validation checkpoint for a batch external operation — the mandatory quiz step ('Do not publish until they confirm') — plus a trust-boundary stop rule before acting on untrusted PRD content and a defined handoff to /implement. | 5 / 5 |
Progressive Disclosure | Sections are well organized and detail is deferred one level deep to clearly signaled external files (root AGENTS.md for the tier definitions, docs/domain-glossary.md for the domain language, scripts/agents.config.sh for tier/domain mappings); the minor gap is that the tier decision procedure is reproduced inline alongside the pointer to its canonical definition. | 4 / 5 |
Total | 16 / 20 Passed |