Install Lians

One memory API, five native SDKs, and a self-hostable server. Pick your language — every client speaks the same REST API (recall, point-in-time recall_at, snapshot, backtest, crypto-shred erasure, audit-chain verify, relationship graph).

Just want to try it? The fastest path is Python local mode — no server, no Docker, no API key:
pip install lians-sdk[local]

Install matrix

Language	Install	Import / entry point
Python	`pip install lians-sdk`	`from lians import LiansClient`
Python (local, no server)	`pip install lians-sdk[local]`	`from lians import LocalLiansClient`
TypeScript / Node	`npm install @lians-ai/lians`	`import { LiansClient } from "@lians-ai/lians"`
Go	`go get github.com/Lians-ai/Lians/agentmem/sdk/go@v0.4.0`	`lians.NewClient(url, key)`
Java (JVM 11+)	Maven `ai.lians:lians-sdk:0.4.0`	`new LiansClient(opts)`
C (C99 + libcurl)	`cmake -B build && cmake --build build`	`lians_client_new(...)`

All SDKs are released in lock-step at the same version (currently 0.4.0).

Run the server

The SDKs (except Python local mode) talk to a Lians server. Self-host it:

git clone https://github.com/Lians-ai/Lians
cd Lians
docker compose up --build        # Postgres + pgvector + API on :8000

Health check: curl localhost:8000/livez. See DEPLOY.md for production (KMS, non-superuser DB role for RLS, air-gap mode, WORM storage).

30-second hello, by language

Python (local — no server)

from lians import LocalLiansClient
from datetime import datetime, timezone

mem = LocalLiansClient()
mem.add(agent_id="desk", content="NVDA guidance raised to $40B",
        event_time=datetime(2025, 11, 19, tzinfo=timezone.utc),
        metadata={"ticker": "NVDA", "metric": "guidance"})
print(mem.recall(agent_id="desk", query="NVDA guidance")["memories"])

Python (server)

from lians import LiansClient
mem = LiansClient(base_url="https://mem.yourfirm.internal", api_key="...")

TypeScript / Node

import { LiansClient } from "@lians-ai/lians";
const client = new LiansClient({ baseUrl: "http://localhost:8000", apiKey: process.env.LIANS_API_KEY! });
await client.addMemory({ agent_id: "desk", content: "NVDA guidance raised to $40B",
                         event_time: "2025-11-19T16:00:00Z", metadata: { ticker: "NVDA" } });
const { memories } = await client.recall({ agent_id: "desk", query: "NVDA guidance" });

Go

import "github.com/Lians-ai/Lians/agentmem/sdk/go"

c := lians.NewClient("http://localhost:8000", os.Getenv("LIANS_API_KEY"))
_, _ = c.AddMemory(ctx, lians.AddMemoryRequest{
    AgentID: "desk", Content: "NVDA guidance raised to $40B",
    EventTime: time.Date(2025, 11, 19, 16, 0, 0, 0, time.UTC),
    Metadata: map[string]any{"ticker": "NVDA"},
})

Java

<dependency>
  <groupId>ai.lians</groupId>
  <artifactId>lians-sdk</artifactId>
  <version>0.4.0</version>
</dependency>

var client = new LiansClient(LiansOptions.builder()
        .baseUrl("http://localhost:8000").apiKey(System.getenv("LIANS_API_KEY")).build());

C

cd agentmem/sdk/c && cmake -B build && cmake --build build   # needs libcurl

#include "lians.h"
lians_client *c = lians_client_new("http://localhost:8000", getenv("LIANS_API_KEY"));

MCP — use Lians as a native tool

Any MCP host (Claude Desktop, Cursor, Windsurf) can use Lians directly. See the MCP section of the README.

Framework integrations (Python)

pip install lians-sdk[langchain]    # LangChain chat history & tools
pip install lians-sdk[langgraph]    # LangGraph node factories
pip install lians-sdk[crewai]       # CrewAI tools
pip install lians-sdk[all]          # everything

Verify a release

pip install lians-sdk==0.4.0
npm view @lians-ai/lians version
go list -m github.com/Lians-ai/Lians/agentmem/sdk/go@v0.4.0

Maintainers: see RELEASING.md — one vX.Y.Z tag ships all five.

FilesExpand file tree

install.md

Latest commit

History