pub struct StoreMemoryResponse {
pub source_id: String,
pub chunks_created: usize,
pub memory_type: String,
pub entity_id: Option<String>,
pub quality: Option<String>,
pub warnings: Vec<String>,
pub extraction_method: String,
pub enrichment: String,
pub hint: String,
}Fields§
§source_id: String§chunks_created: usize§memory_type: StringMemory type at the moment of persistence. If caller did not supply
one and enrichment is pending, this is a placeholder ("fact") —
check enrichment field to know whether to expect it to change.
entity_id: Option<String>§quality: Option<String>§warnings: Vec<String>Schema-validation issues — actionable by the agent.
extraction_method: StringHow structured fields were populated. “agent” | “llm” | “none” | “unknown” (forward-compat default).
enrichment: StringEnrichment state for the memory. "pending" when background
classification + entity extraction + concept linking will run;
"not_needed" when no LLM is available and the memory stays as
caller-supplied. Machine-readable — Tauri app uses this to drive
polling / live-update UI, MCP callers can choose to relay state.
Defaulted for backward compatibility with pre-async-enrichment clients.
hint: StringProse cue for caller agents — safe to relay to the user verbatim.
Communicates that Wenlan is compiling the memory into reusable
context in the background, so callers don’t treat None enriched
fields as failure. Empty when the store completed fully sync.