# Granoflow

Hooks an AI agent up to the local Granoflow app: task tracking, end-of-task reviews, long-term work memory. Useful when an agent should carry past findings forward.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI
- Author: granoflow
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/granoflow/granoflow-mcp-server
- Updated: 2026-07-28
- HTML: https://affpapa.org/en/nevibecode/mcp/granoflow

## What it does

The server stays thin on purpose. It holds no business logic and touches no database of its own. Instead it resolves the local HTTP endpoint of the running Granoflow app, passes structured calls through, and hands back MCP tool results. Task and memory truth lives in the app.

You need Node.js 20 or newer and Granoflow running with the Local HTTP API switched on. The default address is `http://127.0.0.1:56789`; set `GRANOFLOW_API_BASE_URL` to point elsewhere. Access tokens belong in `GRANOFLOW_API_TOKEN` inside your MCP client environment, never in `~/.config/granoflow-mcp/config.json`. Because an env var wins over the saved file, setup reports the override rather than claiming the stored value is live.

Local Granoflow features cost nothing, and without a membership nothing is uploaded. Do not expect this to analyze source or repair CI: tests, linters and engineering judgement stay outside it. Bundled skills cover wrapping up a task, drafting reviews and pulling data in from other agents.

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

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add mcp-server -- npx -y @granoflow/mcp-server
```

### Codex

**Platform CLI**
```bash
codex mcp add granoflow -- npx -y @granoflow/mcp-server
```

### Cursor

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

### GitHub Copilot

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

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add granoflow npx -y @granoflow/mcp-server
```

## When to use it
- record which task an agent is working on
- keep a past session's lesson as a memory card
- draft a weekly review from finished tasks
- import history from Cursor or Codex into the local app
- set a non-default API port without hand-editing config

## Example prompts
- Preview `granoflow_setup_write_config` with the API port for this project
- Fetch the effective agent preferences for this project with `granoflow_agent_preferences_get`
- Draft today's review and write it back to Granoflow

## Similar
- [action-converter](https://affpapa.org/en/nevibecode/skills/action-converter.md) — skill · ★ 783
- [agent-recall](https://affpapa.org/en/nevibecode/mcp/agent-recall.md) — MCP server · ★ 372
- [AccInt](https://affpapa.org/en/nevibecode/mcp/accint.md) — MCP server · ★ 8
- [Mnemosyne OS](https://affpapa.org/en/nevibecode/mcp/mnemosyne-os.md) — MCP server · ★ 20
- [stellavault](https://affpapa.org/en/nevibecode/mcp/stellavault.md) — MCP server · ★ 5
- [session-history](https://affpapa.org/en/nevibecode/skills/session-history.md) — skill · 13900 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