Enum LlmProvider 

pub enum LlmProvider {
    Ollama,
    OpenAiCompatible {
        api_key: String,
    },
}

LLM-provider wire-shape selector. Owned by OllamaClient (the historical name preserved post-#1066 for call-site backward compatibility — a future rename to LlmClient is non-breaking and tracked separately).

#1258 — api_key carries a vendor Bearer token; the manual Drop impl below zeroizes the in-memory bytes when an LlmProvider falls out of scope so the secret does not linger on the heap. #1262 — the manual Debug impl redacts the api_key to <redacted> so a stray {:?} print never leaks the secret.

Variants

Ollama

Ollama native API: POST /api/chat, POST /api/embed. No auth header. This is the historical pre-#1066 behavior and remains the v0.7.0 default.

OpenAiCompatible

OpenAI-compatible API: POST /v1/chat/completions, POST /v1/embeddings. Authorization: Bearer <api_key> header. Covers xAI Grok, OpenAI, Anthropic (via OpenAI shim), Google Gemini, DeepSeek, Kimi, Qwen, Mistral, Groq, Together, Cerebras, OpenRouter, Fireworks, LMStudio, vLLM, llama.cpp server, and any other vendor following the spec.

Fields

api_key: String

Implementations

impl LlmProvider

pub fn zeroize_secrets(&mut self)

#1258 — zeroize the api_key buffer in place. Idempotent. The Drop impl below delegates here so the helper is the single source of truth for the zero-on-secret-loss contract. Tests probe the buffer via this entry point so they observe the post-zeroize state of a still-live allocation (probing after the owning value is dropped is UB — the allocator’s free-list bookkeeping stamps the first 8-16 bytes of the just-freed slot and that’s not a zeroize defect; see #1321).

Trait Implementations

impl Clone for LlmProvider

fn clone(&self) -> LlmProvider

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LlmProvider

fn fmt(&self, f: &mut Formatter<'_>) -> Result

#1262 — redact api_key so accidental {:?} prints never leak the vendor Bearer token. The variant name stays for operator diagnostics.

impl Drop for LlmProvider

fn drop(&mut self)

#1258 — zeroize the api_key buffer on scope exit so the vendor Bearer token does not linger on the heap after the LlmProvider is dropped. Ollama carries no secret and is a no-op. Delegates to LlmProvider::zeroize_secrets.

fn pin_drop(self: Pin<&mut Self>)

🔬This is a nightly-only experimental API. (pin_ergonomics)

Execute the destructor for this type, but different to Drop::drop, it requires self to be pinned.