looot: free search of the data API catalog for up to 5 eligible endpoints by task, capability, category, provider, endpoint, input-schema, output-schema, schema, or hybrid mode (auto infers one). Each result has bounded relevance/evidence/availability/price/freshness components and its category; unknown evidence stays distinct from zero. minimumScore is applied before resultLimit. Identity modes never substitute another endpoint or provider. Never returns a mock endpoint. `detail` changes only how much of each candidate comes back, never ranking or matching: "full" (default) is the whole candidate; "title" returns rank, endpointId, provider, name, capability, category, estimatedPrice, priceBasis, costPerSuccessUsd, async, works {rate, runs, p50Ms, thin} and sourceCapability, enough to pick one to inspect. `capability` is the job id; `sourceCapability` the raw registered slug. Full candidates carry `credential` ("platform_own_key_optional": the platform key runs it, your own key is optional). No endpoint for the named job: candidates is empty and meta.warnings has "no_supply_for_job: <jobId>: ...". maxPrice removed every match: meta.warnings has "price_above_max: <jobId>: ..." naming the cheapest. `jobInputs` are the inputs for the `job:<id>` run; a pinned tool id takes that tool's own field names, so inspect it first. For more than 5 results or a whole category, use `search` (limit up to 200, category/provider/capability/price filters). This search is lexical: when the job is easier to describe in a sentence and nothing here is a confident match, discover_smart judges a shortlist with one small paid AI call (see its description for the cost). `prefer` (balanced default, cheapest, reliable, fastest; `ranking` is an alias) orders providers inside a job; each full candidate's `facts` says why. Field details: https://api.looot.ai/llms-full.txt.