Does Agent Brain send my code or memories to a cloud service?

No. Agent Brain is self-hosted. Storage, vector search, cache, and local model calls run on your machine or infrastructure.

Which agents can use it?

Claude Code, Codex, Cursor, Gemini, Aider, and custom scripts can connect through MCP or REST.

Introduction

Agent Brain is a self-hosted shared memory layer for AI coding agents. It gives Claude, Codex, Cursor, Gemini, Aider, and custom tools a common place to store project knowledge, retrieve relevant context, and reuse semantically similar answers.

It runs locally beside your Docker stack. Your memories, vectors, cache records, and local model operations stay on your machine or infrastructure.

Core capabilities

Persistent memory for decisions, patterns, errors, preferences, and project context
Semantic cache for similar queries and repeated workflows
Cross-agent context sharing through MCP and REST
Local model extraction, scoring, and conflict checks through Ollama
Memory governance for audits, deduplication, archiving, and consolidation
Desktop app for stack lifecycle, dashboard, knowledge explorer, and cache monitor

Ports

| Service | URL | | ------------------- | ------------------------------ | | MCP server | http://localhost:3100/mcp | | Brain Engine REST | http://localhost:9090/api/v1 | | Brain Engine health | http://localhost:9090/health | | Qdrant dashboard | http://localhost:6333 | | Ollama | http://localhost:11434 |

Requirements

Docker Desktop or Docker Engine with Compose
8 GB RAM recommended
macOS 12+, Windows 10+, or Linux x86-64
Enough disk space for local models and service data