langgraph-singlestore

SingleStore-backed persistence for LangGraph: long-term memory (BaseStore) and, in progress, graph checkpointing (BaseCheckpointSaver).

Status

Installation

pip install langgraph-singlestore

Store quickstart

from langgraph.store.singlestore import SingleStoreStore

store = SingleStoreStore(
    host="127.0.0.1",
    port=3306,
    user="root",
    password="",
    database="langgraph",
)
store.setup()  # idempotent; applies pending migrations

store.put(("users", "u1"), key="profile", value={"name": "Ada"})
item = store.get(("users", "u1"), key="profile")

Callers may pass any one of:

• an existing singlestoredb.Connection via connection=...,
• an existing SQLAlchemy Pool via connection_pool=..., or
• plain connection kwargs (host, user, ...) — an internal QueueConnectionPool is built lazily.

connection and connection_pool are mutually exclusive.

TTL

Pass a TTLConfig and start the background sweeper to expire items:

from langgraph.store.base import TTLConfig

store = SingleStoreStore(ttl_config=TTLConfig(...), **conn_kwargs)
store.setup()
store.start_ttl_sweeper()
# ...
store.stop_ttl_sweeper()

Vector search

Pass an index config with dims, embed, and an ann_index_config to enable semantic search():

from langgraph.store.singlestore import SingleStoreStore
from singlestore_langchain_core import ANNIndexConfig

store = SingleStoreStore(
    index={
        "dims": 1536,
        "embed": my_embeddings,
        "ann_index_config": ANNIndexConfig(...),
    },
    **conn_kwargs,
)
store.setup()
results = store.search(("docs",), query="what is singlestore?")

Async

AsyncSingleStoreStore exposes the same constructor and shares the sync class's state; async methods delegate to the sync implementation via the default executor.

from langgraph.store.singlestore import AsyncSingleStoreStore

store = AsyncSingleStoreStore(**conn_kwargs)
await store.asetup()
await store.aput(("users", "u1"), key="profile", value={"name": "Ada"})

Checkpoint saver

SingleStoreSaver / AsyncSingleStoreSaver are placeholders today: the constructors accept configuration but every I/O method raises NotImplementedError. Track progress in CHANGELOG.md.

Layout

This package uses PEP 420 namespace packages under langgraph.checkpoint.* and langgraph.store.*, matching the layout used by langgraph-checkpoint-postgres and friends. Shared connection, filter, and index helpers live in singlestore-langchain-core.

Development

make lint
make test                # unit tests, sockets disabled
make integration_tests   # boots a SingleStore container via singlestoredb