# README

**One brain, every agent.**

Gerber is a Model Context Protocol (MCP) server that gives AI coding agents persistent memory across sessions and projects. Notes, tasks, issues, and inter-session messages — with semantic and full-text search.

## Why Gerber?

AI coding agents forget everything between sessions. Gerber solves this:

* **Notes** — Capture knowledge atoms and long-form documents, searchable via E5 embeddings
* **Tasks** — 7-column kanban (inbox to done) with subtasks, priorities, and due dates
* **Issues** — Bug tracking with severity levels and 4-column workflow
* **Messages** — Inter-session bus for context and reminders between projects
* **Search** — Hybrid engine combining semantic similarity and full-text matching

## Works with every agent

Claude Code, Claude Desktop, Gemini CLI, OpenAI Codex, OpenCode, Kilo Code, Cline — and any MCP-compatible client.

## Get started

|                                                                                             |                                                       |
| ------------------------------------------------------------------------------------------- | ----------------------------------------------------- |
| [**Quickstart**](https://docs-gerber.romain-ecarnot.com/getting-started/quickstart)         | Install and use Gerber in 5 minutes                   |
| [**Concepts**](https://docs-gerber.romain-ecarnot.com/getting-started/concepts)             | Understand projects, notes, tasks, issues, and search |
| [**Tools Reference**](https://docs-gerber.romain-ecarnot.com/tools-reference/projects)      | Complete reference for all 26 MCP tools               |
| [**Claude Managed Agent**](https://docs-gerber.romain-ecarnot.com/deployment/managed-agent) | Connect Gerber to Anthropic's managed agent platform  |