Interface: MemoryConfig
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:202
Top-level configuration object for the Memory facade.
All fields are optional; sensible defaults are applied per field.
A minimal {} config is valid and will use a temporary SQLite brain file
with graph + self-improvement enabled.
Properties
connectionString?
optionalconnectionString:string
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:229
Postgres connection string. Required when store='postgres'.
Example
'postgresql://postgres:wunderland@localhost:5432/agent_memory'
consolidation?
optionalconsolidation:ExtendedConsolidationConfig
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:282
Consolidation schedule and thresholds.
decay?
optionaldecay:boolean
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:279
Whether memory traces lose strength over time following an Ebbinghaus forgetting-curve model.
Default
true
embed()?
optionalembed: (text) =>Promise<number[]>
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:258
Optional embedding function for generating vectors at remember/recall time. When provided, enables HNSW vector search in recall() and stores embeddings alongside text in remember(). Without this, recall() falls back to FTS5-only.
Parameters
text
string
Returns
Promise<number[]>
Example
const mem = await Memory.createSqlite({
embed: async (text) => {
const res = await openai.embeddings.create({ model: 'text-embedding-3-small', input: text });
return res.data[0].embedding;
},
});
embeddings?
optionalembeddings:EmbeddingConfig
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:241
Embedding model configuration (provider name + optional model).
forgotten?
optionalforgotten:ForgottenLedger|Omit<ForgottenLedgerOptions,"primary">
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:307
The forgotten-user ledger this store's deletions and guards use: one already open, or the options to open one
over this store's own adapter. Memory.markForgotten and Memory.isForgotten write and read it. A store opened
without one has neither call.
With options, the ledger's table lives in the store's own database, and copy names the adapter of the second
copy kept outside it (openForgottenCopy opens one). The open is refused, and the store closed again, when the
ledger does not open: its copy is missing, or its tombstones are keyed and the key is not given. The open also
compares the ledger identity the store holds with its copy's, and refuses a copy that belongs to another store.
Tombstones are database-wide, including historical learner rows: brain ids do not isolate them. Give each agent its own database or PostgreSQL schema and a separate copy outside restores.
Example
const { adapter: copy } = await openForgottenCopy('./data/forgotten.sqlite');
const mem = await Memory.createSqlite('./data/agent-memory.sqlite', { forgotten: { copy } });
await mem.markForgotten({ kind: 'user', id: userId });
graph?
optionalgraph:boolean
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:265
Whether to build and maintain a knowledge graph alongside the vector store. When enabled, entity co-occurrence and semantic edges are tracked.
Default
false
ingestion?
optionalingestion:IngestionConfig
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:285
Document ingestion settings applied to all ingest() calls by default.
path?
optionalpath:string
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:223
File-system path for stores that require one (e.g. SQLite db file). Ignored by in-memory and remote stores.
Example
'./data/agent-memory.sqlite'
qdrantApiKey?
optionalqdrantApiKey:string
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:238
Qdrant API key for cloud instances. Optional.
qdrantUrl?
optionalqdrantUrl:string
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:235
Qdrant base URL. Required when store='qdrant'.
Example
'http://localhost:6333'
selfImprove?
optionalselfImprove:boolean
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:272
Whether the agent may autonomously refine and restructure its own memories (write new insight traces, prune contradictions, merge redundancies).
Default
false
store?
optionalstore:"memory"|"postgres"|"qdrant"|"sqlite"|"neo4j"|"hnsw"
Defined in: packages/agentos/src/cognition/memory/io/facade/types.ts:216
Persistence backend for memory traces.
The Phase 1 facade currently implements the SQLite path at runtime. Other values are reserved for future backends and will throw if selected.
'sqlite'– file-based SQLite (implemented; recommended).'memory'– reserved for a future in-process backend.'qdrant'– reserved for a future vector-database backend.'neo4j'– reserved for a future graph-database backend.'hnsw'– reserved for a future ANN-only backend.
Default
'sqlite'