Content
93%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 high-quality routing skill: dense with vendor-specific operational facts, exact endpoints and constraints, explicit confirmation gates around production writes, and well-structured one-level-deep delegation to documentation and sub-skills. The only gap is that error-recovery/verification loops live in the linked sub-skills rather than being summarized in the body.
Suggestions
Add a brief error-recovery note to the body (e.g., what to check when an ingest call returns non-200, or how to verify a contract edit took effect via a GET before informing the user), so the main file carries a minimal validate-loop the way it carries the pre-write confirmation gates.
The routing table and the 'Key documentation' section both surface the llms.txt index and the Stripe integration guide; merging the duplicated links would trim a few tokens without losing any navigation value.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every line carries non-obvious Metronome facts Claude cannot know — "24 hours after billing_period_end_date by default", "250 line items per Stripe invoice", "1000 is 10.00 USD", "a bare JSON array of 1 to 100 event objects" — with zero generic concept explanation. It assumes competence and every token earns its place, matching the lean level-5 anchor. | 5 / 5 |
Actionability | Guidance is fully executable: exact endpoints ("POST /v1/ingest", "POST /v2/contracts/edit", "POST /v1/credits/migrateToContracts"), exact field names (transaction_id, usage_filter, rollover_fraction, applicable_product_ids), exact body shapes, and hard numeric limits. As an instruction-only skill its concrete, specific guidance satisfies the actionability bar without code. | 5 / 5 |
Workflow Clarity | The route → read-linked-page → confirm → execute sequence is clear, with explicit pre-write validation gates ("get the user's explicit confirmation before any write", "identify the Stripe account and mode behind the Stripe key… and confirm those writes separately"). However, explicit validate/retry feedback loops for error recovery are delegated to the linked sub-skills rather than present in the body itself, so it fits "most checkpoints present; minor validation gaps" rather than the level-5 anchor requiring feedback loops and checklists. | 4 / 5 |
Progressive Disclosure | The body is a clean overview router: two well-organized tables map each task to a one-level-deep linked doc or hosted skill, with a clearly signaled fallback (the llms.txt index and "append .md to the page URL"). No local bundle files exist, and content is appropriately split out — the inline Critical rules are exactly the high-frequency safety/correctness facts that belong in the main file. | 5 / 5 |
Total | 19 / 20 Passed |