Struct LlmClient 

pub struct LlmClient { /* private fields */ }

Implementations

impl LlmClient

pub fn new( base_url: Option<&str>, model: &str, key: Option<&str>, ) -> Result<LlmClient, Error>

A client for model at base_url (OpenAI when absent), with key.

pub fn token_usage(&self) -> (u64, u64)

Cumulative (input, output) tokens used so far, across all calls.

pub fn with_models(self, models: BTreeMap<Task, String>) -> LlmClient

Run task on its own model instead of the default one.

pub fn model(&self, task: Task) -> &str

The model task runs on.

pub async fn enrich_symbol( &self, prompts: &Prompts, brief: &str, ) -> Result<Enrichment, Error>

One-line summary of what a symbol does, plus a few natural-language queries it answers.

pub async fn guide_section( &self, prompts: &Prompts, repo: &str, title: &str, guidance: &str, evidence: &str, ) -> Result<Option<String>, Error>

ONE section of a repo’s style guide from its tailored evidence, as bullet rules without a heading, or None when the evidence shows no consistent convention.

pub async fn narrate_flow( &self, prompts: &Prompts, entry: &str, signature: &str, trace: &str, ) -> Result<(String, Vec<String>), Error>

Narrate an execution flow: given an entry point and its ordered call trace, produce a concise markdown walkthrough plus a few flow-oriented questions the flow answers.