Ambient Memory: an open-source local memory layer for AI agents, by Jarvis Osorio

Jarvis Osorio (Jarvis Nelson Osorio) has open-sourced Ambient Memory, a local-first context layer for AI engineers. It passively remembers what you worked on — files, screen sessions and browser history — and exposes that memory to Cursor and Claude Desktop over the Model Context Protocol (MCP).

Project page: https://jarvisosorio.com/ambient-memory/

The repository is TypeScript, requires Node.js 20 or later (22+ preferred), and is released under the MIT License. The current version is 0.3.0. There is no telemetry or network sync in v0.3.

What it does

A background daemon watches Documents, Desktop, Downloads and the current working directory, and indexes screen and browser activity with noise filters. Only the daemon opens the SQLite vault. The CLI and the MCP server are thin clients over a local Unix socket (or a Windows named pipe).

Storage uses Node's built-in node:sqlite on Node 22+, with a sql.js WASM fallback on Node 20. There is no better-sqlite3 native addon and no node-gyp install.

Search uses FTS5/BM25. Entity fields are encrypted with AES-256-GCM. When the daemon stops, the database is sealed to memory.db.enc. While the vault is unlocked, the FTS index is redacted plaintext — an intentional search tradeoff, documented in the repo.

MCP tools

memory_search, memory_recent, memory_get, memory_context, memory_relate, and memory_status.

Quick start

As of 12 November 2026 the package is not yet on the npm registry, so install from GitHub:

git clone https://github.com/JarvisOsorio/ambient-memory.git
cd ambient-memory
npm install
npm test
node ./bin/ambient.js init
node ./bin/ambient.js start --screen auto
node ./bin/ambient.js mcp

Once the package is published to npm, the README's intended install is npx -y ambient-memory init (and the same for start and mcp).

Jarvis Osorio is a technology entrepreneur and AI innovator based in Miami Beach, Florida. Official site: https://jarvisosorio.com/