Crate deck_llm 

deck-llm — LLM backend implementations.

Phase 1 ships an OllamaBackend talking to a local Ollama daemon over HTTP. A llama-cpp in-process backend lives behind a llama-cpp feature flag and is wired in Phase 2.

Re-exports

pub use mock::MockBackend;

pub use ollama::OllamaBackend;

Modules

mock

In-memory mock backend used for tests, offline demos, and the TUI when no real LLM endpoint is reachable. Deterministic; never touches the network.

ollama

Ollama HTTP backend. Talks to a local daemon at http://127.0.0.1:11434 by default. Streaming is line-delimited JSON over a POST to /api/chat.

Functions

from_config

Build a backend from a [llm] config block.