resolve_skill

What it does

`resolve_skill` is the discovery step. Call it when the agent knows the task it needs help with but has not chosen the exact Milkey skill yet.

The result should be treated as selection guidance. After you get the best match, the next step is usually to call `get_skill` for the chosen slug.

When to use it

• The user asked for guidance in a domain with multiple possible skills, such as frontend, backend, database, or deployment work.
• You want the agent to choose a specific skill intentionally instead of guessing or hardcoding a slug.
• You need category-aware routing, such as preferring a frontend skill when the request could plausibly match multiple domains.

Inputs

• query (required): A short, high-signal request such as `nextjs docs sidebar content structure` or `fastapi jwt auth`.
• category (optional): An optional preference like `frontend`, `backend`, or `data` when you want tighter routing.

1{2  "query": "nextjs docs sidebar content structure",3  "category": "frontend"4}

Response shape

• The best matching skill slug plus supporting metadata about why it matched.
• Related skills when the task is broad and the user may need a nearby alternative.
• Enough information to decide whether to fetch the winning skill body immediately.

In practice, the most important handling rule is:

• Use the top-ranked slug first unless the response clearly signals ambiguity.
• When the task is broad, keep the related skills around as fallback options instead of rerunning discovery immediately.

After resolve_skill, the next page to read is get_skill.