@juspay/neurolink
Version:
The pipe layer of an AI nervous system: one interface connecting provider neurons to your application, across three inference types — generate, stream and decide. `decide` returns typed, calibrated judgements from a non-generative model (~400ms, ~$0.00002
11 lines (10 loc) • 450 B
JavaScript
/**
* Tool resolution types — the single tool-policy pipeline that decides which
* tools are sent to a provider for a given request.
*
* One policy object is resolved per request from (a) per-call legacy options
* (`toolFilter`, `excludeTools`, `enabledToolNames`, `disableTools`) and
* (b) the instance-level `tools` config, then applied at the single gate in
* `BaseProvider` that every generate/stream path passes through.
*/
export {};