Expand description
Axum-backed in-process mock LLM server.
MockLlm serves one provider endpoint from a caller-supplied responder
closure. Per request it records facts useful for assertions (call count,
tool declarations, tool_call_ids of returned tool results) into a
shared MockState, derives RequestFacts, and returns whatever the
responder builds (typically via crate::shapes).
Structs§
- Mock
Handle - Handle to a spawned mock server.
- MockLlm
- Builder for an in-process mock LLM server.
- Mock
State - Shared request-fact recorder. Cheap to clone; all clones share state.
- Request
Facts - Facts derived from one incoming request, handed to the responder.
Enums§
- Provider
- Which provider wire the mock speaks.