crewai-persistence-sql

SQL persistence backend for CrewAI Flows via SQLAlchemy — one provider for SQLite, PostgreSQL, MySQL, and more, selected by connection URL. Optional built-in message pruning via agentstate-reducer. Implements CrewAI's FlowPersistence interface.

Installation

pip install crewai-persistence-sql               # SQLite built-in
pip install "crewai-persistence-sql[postgres]"   # + psycopg
pip install "crewai-persistence-sql[mysql]"      # + pymysql
pip install "crewai-persistence-sql[reducer]"    # + message pruning

Requires Python 3.10–3.13.

Usage

from crewai.flow.flow import Flow, start
from crewai.flow.persistence import persist
from crewai_persistence_sql import SQLFlowPersistence

backend = SQLFlowPersistence(url="sqlite:///crewai_flows.db")
# or  "postgresql+psycopg://user:pass@host/db"  /  "mysql+pymysql://user:pass@host/db"

@persist(backend)
class MyFlow(Flow[MyState]):
    @start()
    def begin(self):
        self.state.counter += 1

With message pruning:

from agentstate_reducer import MessageReducer
backend = SQLFlowPersistence(
    url="sqlite:///crewai_flows.db",
    reducer=MessageReducer(min_messages=10, max_messages=20),
    messages_key="messages",
)

API

SQLFlowPersistence(url=None, *, table_name="crewai_flow_states", engine=None, reducer=None, messages_key="messages")

Data model

One row per flow: flow_uuid (primary key) + data (the flow state as a JSON string). Each save upserts (portable update-then-insert), so load_state returns the latest.

License

MIT

Long-term memory via on_prune (agentstate-reducer >= 0.4.0)

Messages pruned from the flow state are exactly the ones leaving the model's view. The persistence layer forwards a memory namespace to the reducer, and any on_prune hook receives (pruned_messages, namespace) — so pruned turns can flow straight into CrewAI's unified Memory (or any store), with no package coupling:

from agentstate_reducer import MessageReducer, ReducerConfig, Background
from crewai.memory import Memory

memory = Memory(storage=...)                       # e.g. crewai-memory-dynamodb

def remember(pruned, namespace):
    text = "\n".join(m["content"] for m in pruned)
    memory.remember_many(memory.extract_memories(text), scope=namespace)   # LLM calls -> run off-path

reducer = MessageReducer(config=ReducerConfig(max_messages=20, on_prune=[Background(remember)]))

class SupportState(BaseModel):
    id: str = ""
    memory_namespace: str = "/user/kamal"          # long-term scope: the USER, not the flow
    messages: list = []

@persist(SQLFlowPersistence(..., reducer=reducer))
class SupportFlow(Flow[SupportState]):
    ...

The persistence layer reads memory_namespace (or whatever ReducerConfig.namespace_key names) from the flow state on every save_state and passes it through untouched. If the state never sets it, the namespace falls back to "/flow/<flow_uuid>". Each pruned message reaches the hooks once.