Struct AgentConfig 

pub struct AgentConfig {
    pub db_path: PathBuf,
    pub sessions_dir: PathBuf,
    pub hints_dir: PathBuf,
    pub acceptance_window_minutes: i64,
    pub score_recompute_interval_secs: u64,
    pub top_k: usize,
    pub classifier: Option<Arc<dyn TaskClassifier>>,
}

Runtime config for the agent loop.

Fields

db_path: PathBuf

SQLite path. Default: default_db_path() (caller-provided).

sessions_dir: PathBuf

Root containing <dir>/<session>.jsonl files. Default: ~/.claude/projects.

hints_dir: PathBuf

Where to write <session>.md hint files. Default: ~/.claude/hints.

acceptance_window_minutes: i64

How long after a suggestion a matching tool_use still counts as “accepted”. Default: 60 minutes.

score_recompute_interval_secs: u64

How often the engine recomputes tool_scores (seconds).

top_k: usize

Number of recommendations to write into the hint file.

classifier: Option<Arc<dyn TaskClassifier>>

Optional task classifier. None ⇒ raw user text is sent to the recommender verbatim (default). When set, every UserText event is piped through the classifier first; non-task messages are dropped and real tasks are rewritten into a focused query before embedding.

Implementations

impl AgentConfig

pub fn new(db_path: PathBuf, sessions_dir: PathBuf, hints_dir: PathBuf) -> Self

pub fn with_classifier(self, classifier: Arc<dyn TaskClassifier>) -> Self

Trait Implementations

impl Clone for AgentConfig

fn clone(&self) -> AgentConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.