RAG Context Adapter

Augment context with retrieved documents.

RagContextAdapter implements ContextAdapter by searching an EmbeddingStore for documents relevant to the user's input, then prepending them as system messages.

The full file is src/rag/mod.rs.

Backends

• Qdrant — feature qdrant. Vector search.
• Tantivy — feature tantivy. Full-text search.

API

pub struct RagContextAdapter {
    store: Arc<dyn EmbeddingStore>,
    config: RagConfig,
}

pub struct RagConfig {
    pub max_chunks: usize,        // default 5
    pub chunk_prefix: String,     // default "Relevant context:"
    pub min_score: f32,           // default 0.7
}

See also

• ContextAdapter — the trait.
• EmbeddingStore — the backend.