# crbro-memory

A local MCP server that keeps an agent's knowledge across sessions: facts, decisions, errors and session summaries stored as files.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 8
- Author: Octonove
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/Octonove/crbro-memory
- Updated: 2026-10-18
- HTML: https://affpapa.org/en/nevibecode/mcp/crbro-memory

## What it does

CRBRO splits knowledge into neurons (cortex), links between them (synapses) and session memory (hippocampus). Search runs over individual facts through Orama, so a topic holding hundreds of entries stays as findable as a small one. Every hit returns the matching line, the date it was written and a `confidence` label.

You can read memory in slices. `crbro_inspect view=neuron` gives an index of facts, decisions, patterns and errors with an id, a date and a preview, while `entries=[ids]` pulls only those. Facts carry a shelf life — `volatile`, `normal`, `durable` or `permanent`, set through `CRBRO_SHELF_DAYS`; expired claims move to `possibly_stale` with a source-check warning. Repeating a fact does not verify it: only `crbro_revise status=verified` restarts a check.

\`npx crbro-memory init` installs a local embedding model, `multilingual-e5-small` in int8, and fuses it with keyword search. That costs roughly 500 MB on disk once per machine and about 0.5 GB of RAM while a server runs. `init --no-semantic` skips the install, and `CRBRO_SEMANTIC=0` turns semantics off later. Node.js is required, since the server starts through `npx`.

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

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add crbro-memory -- npx -y crbro-memory
```

### Codex

**Platform CLI**
```bash
codex mcp add crbro-memory -- npx -y crbro-memory
```

### Cursor

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

### GitHub Copilot

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

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add crbro-memory npx -y crbro-memory
```

## When to use it
- keep decisions and agreements alive between agent sessions
- find one fact in a topic that already holds hundreds
- separate stale claims from current ones
- read only the entries you need instead of a whole neuron
- record whether a claim came from a source or from the user

## Example prompts
- Save to memory that we moved hosting to Hetzner
- Search memory for everything about deploys from the last two weeks
- Show the project neuron index and read only the migration entries

## Similar
- [REA](https://affpapa.org/en/nevibecode/mcp/rea.md) — MCP server · ★ 386989
- [GitHub](https://affpapa.org/en/nevibecode/mcp/github.md) — MCP server · ★ 33918
- [Netdata](https://affpapa.org/en/nevibecode/mcp/netdata.md) — MCP server · ★ 81115
- [browser-use](https://affpapa.org/en/nevibecode/mcp/browser-use.md) — MCP server · ★ 119160
- [Codebase Memory](https://affpapa.org/en/nevibecode/mcp/codebase-memory.md) — MCP server · ★ 47392
- [mcp-server](https://affpapa.org/en/nevibecode/mcp/mcp-server.md) — MCP server · ★ 43872

---
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