AgentRecall

Plug-and-play persistent memory for AI agents. Open source, zero config.

Python · Node.js · Local-first · Semantic search · Auto-compression · Confidence decay

Packages

Quick Start

Python:

from agentrecall import MemoryStore

store = MemoryStore()
store.remember("User prefers dark mode", agent="assistant")
memories = store.recall("What theme should I use?", agent="assistant")

Node.js:

import { MemoryStore } from "agentrecall";

const store = new MemoryStore();
await store.remember("User prefers dark mode", { agent: "assistant" });
const memories = await store.recall("What theme should I use?", { agent: "assistant" });

Cloud Mode

Both SDKs support a cloud mode that connects to the AgentRecall Cloud API:

Python:

from agentrecall import CloudClient, AgentMemoryConfig

config = AgentMemoryConfig(cloud_url="https://api.agentrecall.cloud", api_key="your-key")
client = CloudClient(config)
client.remember("User prefers dark mode", agent="assistant")
memories = client.recall("What theme should I use?", agent="assistant")

Node.js:

import { CloudClient } from "agentrecall";

const client = new CloudClient("https://api.agentrecall.cloud", "your-key");
await client.remember("User prefers dark mode", { agent: "assistant" });
const memories = await client.recall("What theme should I use?", { agent: "assistant" });

Features

• Semantic Recall — hybrid scoring with embeddings (optional)
• Auto-Classification — memories tagged as preferences, corrections, facts, temporal
• Confidence Decay — memories fade unless reinforced
• Skip/Penalty — mark irrelevant memories to reduce recall score
• SQLite — zero config, works offline, single file
• Privacy First — runs locally, no data leaves your machine

Pricing

Contributing

1. Fork the repo
2. Pick a package (packages/python or packages/node)
3. Make your changes
4. Run tests:

• Python: cd packages/python && python -m pytest tests/
• Node.js: cd packages/node && npm test

1. Open a PR

License

MIT