Quickstart — Python

The Python SDK is generated from the same OpenAPI document the TypeScript SDK uses, so the surface is identical and the methods feel native to Python.

Run Mnemosyne (one container) and install the SDK

docker run --rm -d --name mnemo \
  -e DATABASE_URL=postgres://mnemo:mnemo@host.docker.internal:5432/mnemo \
  -e MNEMO_LLM_PROVIDER=openai \
  -e MNEMO_LLM_API_KEY=sk-... \
  -p 3000:3000 ghcr.io/lucasmailland/mnemosyne-server:latest
 
# Mint a workspace key to authenticate the SDK
docker exec mnemo node scripts/create-api-key.cjs --workspace demo
 
pip install mnemosyne-sdk

First call

app.py

from mnemo import MnemosyneClient
 
mnemo = MnemosyneClient(
    base_url="http://localhost:3000",
    api_key="mns_live_demo",
)
 
mnemo.facts.create(content="Lucas prefers espresso", tags=["coffee"])
hits = mnemo.recall(query="coffee preferences", top_k=3)
for hit in hits.hits:
    print(f"{hit.score:.3f}  {hit.statement}")

FastAPI integration

main.py

from fastapi import FastAPI, Depends
from mnemo import MnemosyneClient
 
app = FastAPI()
 
def get_memory():
    return MnemosyneClient(
        base_url="http://localhost:3000",
        api_key="mns_live_demo",
    )
 
@app.post("/agents/recall")
def agent_recall(query: str, mnemo: MnemosyneClient = Depends(get_memory)):
    return mnemo.recall(query=query, top_k=5)

LangChain adapter

from mnemosyne.langchain import MnemosyneMemory
 
memory = MnemosyneMemory(
    client=mnemo,
    namespace="conversation:user_42",
)
# Drop straight into any chain that consumes a BaseMemory

LlamaIndex adapter

from mnemosyne.llamaindex import MnemosyneRetriever
 
retriever = MnemosyneRetriever(client=mnemo, top_k=5)
nodes = retriever.retrieve("what does the user prefer?")

💡

The LangChain and LlamaIndex adapters are thin wrappers — they call the same recall endpoint. If they don’t match your needs, drop them and use MnemosyneClient directly.

Node / TypeScript From scratch