Conciseness

The body is dense with genuinely non-obvious internal OpenMeter knowledge (state machine wiring, tax dual-write, line-engine ownership) that Claude would not already know, so most tokens earn their place. It is not score 3 because the ~565-line body has real redundancy — side-effect hook behavior is restated across the Line Engine Lifecycle section, the App trigger interaction section, and the Non-Obvious Gotchas section — and several passages could be tightened.

Actionability

Provides executable Go snippets (the union-type constructor pattern, the tax code write/read flows), exact file paths and line references (`billing/adapter/stdinvoicelines.go` line 755), and copy-paste-ready checklists ('Adding a new field checklist', 'When adding a new line engine or hook'). This matches fully executable, specific, copy-paste-ready guidance rather than the pseudocode/incomplete score-2 anchor.

Workflow Clarity

Multi-step processes are clearly sequenced with explicit validation checkpoints and feedback loops: the numbered 'Adding a new field checklist', the 5-step 'When adding a new line engine or hook', the customer-locking sequence, and the state machine flow with retryable states (TriggerRetry → DraftValidating) and validation gates (noCriticalValidationErrors). This matches clear sequence with explicit validation and feedback loops, not the score-2 'checkpoints missing or implicit'.

Progressive Disclosure

SKILL.md is a well-sectioned overview with a package map and key-files table, and it offloads detail to two real, clearly-signaled one-level-deep references — 'See `references/subscription-sync.md` for full details' and 'See `references/testing.md` for full test patterns' (both files verified to exist). References are one level deep with easy navigation, matching the well-signaled one-level-deep anchor rather than the score-2 'content that should be separate is inline'.

Specificity

Lists many concrete capabilities and areas — 'invoice lifecycle, billing profiles, customer overrides, invoice line items, gathering invoices, standard invoices, the invoice state machine, billing validation issues, billing-subscription sync, the billing worker, invoice calculation, rating/pricing engine, or tax config' plus integration tests, adapter, HTTP handlers, and exact file paths. This matches the anchor for listing multiple specific concrete actions, not the narrower score-2 'names domain and some actions'.

Completeness

Explicitly answers both 'what' (Work with the OpenMeter billing package + enumerated areas) and 'when' via three explicit trigger clauses: 'Use this skill whenever touching...', 'Also use when writing or debugging...', and 'Trigger this skill for any file under...'. This clearly answers both what AND when with explicit triggers, not the score-2 case where 'when' is only implied.

Trigger Term Quality

Covers natural terms a developer would actually say ('invoice lifecycle', 'billing profiles', 'customer overrides', 'tax config', 'billing integration tests') alongside explicit file-path triggers like `openmeter/billing/` and `test/billing/`. Good coverage of common variations rather than only some relevant keywords (score 2).

Distinctiveness Conflict Risk

Has a clear niche (the OpenMeter billing package) with distinct file-path-gated triggers and even carves out the charges subpackage as a separate skill. Unlikely to trigger for the wrong skill, matching the clear-niche anchor rather than the score-2 'could still overlap with similar skills'.