🧠 AgentRecall

The memory SDK for AI agents. Store, search, and traverse memories with graph-powered intelligence.

What is AgentRecall?

AgentRecall gives your AI agents persistent, searchable memory backed by a graph database. Every memory is automatically processed by AI to extract entities, detect relationships, and build a connected knowledge graph.

• Semantic Search — Find relevant memories with natural language queries
• Graph Memory — Neo4j-powered relationship traversal to uncover hidden connections
• AI Processing — Qwen2.5-7B automatically enriches every memory
• Multi-Agent — Isolate memory per agent or share across agents
• Cloud or Local — Use our hosted API or run everything yourself

Installation

pip install agentrecall-sdk

npm install agentrecall-ai-sdk

Quick Start

Python

from agentrecall_sdk import AgentRecall

# Connect to the cloud API
memory = AgentRecall(
    api_key="ar_your_api_key",
    mode="cloud"
)

# Store a memory
memory.store(
    content="The user prefers dark mode and Python over JavaScript",
    agent_id="my-agent",
    category="preferences"
)

# Search memories
results = memory.search(
    query="what UI preferences does the user have?",
    agent_id="my-agent"
)

for result in results:
    print(result.content, result.score)

Node.js

import { AgentRecall } from "agentrecall-ai-sdk";

const memory = new AgentRecall({
  apiKey: "ar_your_api_key",
  mode: "cloud",
});

// Store a memory
await memory.store({ 
  content: "The user prefers dark mode and Python over JavaScript",
  agentId: "my-agent",
  category: "preferences"
});

// Search memories
const results = await memory.search({
  query: "what UI preferences does the user have?",
  agentId: "my-agent",
});

results.forEach(r => console.log(r.content, r.score));

Cloud vs Local