# Memorix

A local project memory layer for AI agents that keeps decisions, facts, and context available across chats, IDEs, and sessions.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 863
- Author: AVIDS2
- License: Apache-2.0 · policy=link-only · description only; the file stays with the author
- Source: https://github.com/AVIDS2/memorix
- Updated: 2026-10-23
- HTML: https://affpapa.org/en/nevibecode/mcp/memorix

## What it does

Memorix gives an AI coding agent shared memory scoped to a Git project. The stored context remains available after a new chat, an IDE switch, or a different terminal session. It covers searchable facts, fixes, gotchas, session summaries, and implementation notes.

The server also turns Git history into engineering facts about changes and their purpose. Design rationale can include alternatives, trade-offs, and risks. Task context, resume flows, and `memorix orchestrate` support work shared by parallel agents.

SQLite is the main store, while its local indexes can be rebuilt. LLM-based memory formation and embeddings are optional. Code Memory has source-backed symbol and relation data for TypeScript and JavaScript; other languages use a Lite fallback, so code analysis depends on the project language.

## Requirements
- Needs: Node.js (commands run via npx)

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add memorix -- npx -y memorix
```

### Codex

**Platform CLI**
```bash
codex mcp add memorix -- npx -y memorix
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "memorix": {
            "command": "npx",
            "args": [
                "-y",
                "memorix"
            ]
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "memorix": {
            "type": "stdio",
            "command": "npx",
            "args": [
                "-y",
                "memorix"
            ]
        }
    }
}
```

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add memorix npx -y memorix
```

## When to use it
- carry project context across agent sessions
- search earlier fixes and implementation notes
- turn commits into searchable engineering facts
- preserve architecture decisions and trade-offs
- coordinate context between parallel agents

## Example prompts
- Save the decision to use SQLite for local project memory and record its reasons.
- Find earlier fixes related to authentication and show their project sources.
- Prepare context for resuming this task in a new session, including known constraints.

## Similar
- [context-sync](https://affpapa.org/en/nevibecode/mcp/context-sync.md) — MCP server · ★ 189
- [timeline-report](https://affpapa.org/en/nevibecode/skills/timeline-report.md) — skill · 6800 installs
- [pp-espn](https://affpapa.org/en/nevibecode/skills/pp-espn.md) — skill · 3800 installs
- [ontology](https://affpapa.org/en/nevibecode/skills/ontology.md) — skill · 3700 installs
- [pp-movie-goat](https://affpapa.org/en/nevibecode/skills/pp-movie-goat.md) — skill · 3400 installs
- [datapackage](https://affpapa.org/en/nevibecode/skills/datapackage.md) — skill · 2100 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md