Conciseness

The body is lean and assumes Claude's competence: every line is a file path, command, or workflow step, with no explanation of concepts Claude already knows. It even enforces token discipline explicitly ('Do not inspect the full base.json in chat. Use the scripts so the large source model body stays out of context'). Not score 4 because there is no padding to trim — even the alias snippet and usageFactor notes are domain-specific facts Claude could not know.

Actionability

All guidance is copy-paste executable: concrete 'node .opencode/skills/openwork-models/scripts/openwork-models.mjs search "z-ai/glm-5.1"' style commands cover every common case (search, add, remove, discount, sync, validate, trigger workflow), and the validation section gives exact commands to run before finishing. Not score 4 because the commands cover the common cases completely with real arguments, matching the anchor-5 example pattern.

Workflow Clarity

Add/Remove/Discount workflows are clearly sequenced with ambiguity-resolution checkpoints (steps 2-3 of Add Workflow: ask the user, list matching IDs), scripts validate after each mutation, and the Validation section mandates running 'validate' and 'build-models.mjs' before finishing. The Update Models workflow includes explicit failure handling ('If the workflow or PR fails/closes unmerged, report that failure with the run/PR URL'). This satisfies the destructive/batch validation cap — validation and feedback loops are present — so it is not capped at 3 or reduced to 4.

Progressive Disclosure

Scored against the actual bundle: all three scripts referenced in the body (extract-source-models.mjs, openwork-models.mjs, trigger-update-workflow.mjs) exist in ./scripts/, and the referenced subcommands (--query, search/add/remove/discount/sync/validate, --base/--ref) are real. The SKILL.md is a concise overview with heavy data (base.json) deliberately kept out of context behind scripts, and references are one level deep with no nesting. Not score 4 because there are no organization gaps — sections map cleanly to the workflows and the script surface.

Specificity

The description lists multiple concrete actions — 'manage OpenWork inference model aliases, openwork model overlays, discounts, validation, and automated base model refreshes from models.dev' — with comprehensive coverage of the skill's operations (add, remove, discount, audit, update). It is not score 4 because coverage is complete rather than having minor gaps; not below because every action named is concrete rather than generic.

Completeness

It clearly answers both 'what' (manage aliases, overlays, discounts, validation, automated base model refreshes) and 'when' with the explicit clause 'Use when adding, removing, discounting, auditing, or updating OpenWork models, including requests like "update the models"'. It matches the anchor-5 example structure exactly; a 4 would require the 'when' clause to be less explicit or lack concrete trigger phrases.

Trigger Term Quality

Natural trigger terms are comprehensively covered: 'adding, removing, discounting, auditing, or updating OpenWork models' plus the explicitly quoted user phrase "update the models" and the models.dev refresh trigger. Not score 4 because it goes beyond good keyword coverage by including the exact natural phrasing a user would say and synonym variants (update/refresh); no meaningful natural terms are missing for this domain.

Distinctiveness Conflict Risk

The OpenWork inference-model niche is highly specific — aliases, overlays, the update-models GitHub workflow, and models.dev refreshes — with distinct triggers unlikely to fire for any other skill. Not score 4 because there is no identifiable overlap with a closely related skill; the domain name scopes every trigger.