ask-ruvnet-mcp

An MCP server that manages NotebookLM source refresh and studio generation for the Ask-RuvNet ecosystem. It monitors GitHub repositories for changes, syncs sources into NotebookLM, and orchestrates studio artifact creation.

Quick Start

Add the server to Claude Code:

claude mcp add ask-ruvnet -- npx ask-ruvnet-mcp

Or run directly:

npx ask-ruvnet-mcp

Environment Variables

| Variable | Required | Default | Description | |---|---|---|---| | NLM_NOTEBOOK_ID | Yes | — | NotebookLM notebook ID to manage | | NLM_BIN_PATH | No | nlm | Path to the nlm CLI binary | | NLM_ASSETS_DIR | No | ./assets | Directory for downloaded studio artifacts | | NLM_REGISTRY_PATH | No | ./nlm-url-registry.json | Path to the source URL registry JSON | | NLM_STATE_PATH | No | ./.nlm-refresh-state.json | Path to the auto-managed state file | | GITHUB_TOKEN | No | — | GitHub token for accessing private repositories |

Tools

| Tool | Description | |---|---| | nlm_refresh_sources | Check GitHub repos for new commits and refresh stale NotebookLM sources. Supports --dry-run, --force, and single-source refresh. | | nlm_studio_create | Create a studio artifact: audio, video, infographic, or slides. Accepts an optional focus prompt. | | nlm_studio_status | Check generation status for a specific studio or all studios. | | nlm_studio_download | Download a completed studio artifact to the assets directory. | | nlm_pipeline_status | Get pipeline health: last refresh time, pending changes, auth status, and studio counts. | | nlm_registry_list | List all tracked source URLs with sync status, last SHA, and enabled/disabled state. |

Usage Examples

Refresh sources (dry run):

{
  "name": "nlm_refresh_sources",
  "arguments": { "dry_run": true }
}

Create an audio overview with a focus prompt:

{
  "name": "nlm_studio_create",
  "arguments": { "type": "audio", "focus": "Summarize the latest architecture changes" }
}

Download a completed studio artifact:

{
  "name": "nlm_studio_download",
  "arguments": { "studio_id": "abc123", "type": "audio" }
}

Prerequisites

1. Node.js 20+ is required.

2. notebooklm-cli must be installed and authenticated:

   npm install -g notebooklm-cli
   nlm login

3. Set NLM_NOTEBOOK_ID to the target notebook before starting the server.

Protocol

This server implements the Model Context Protocol (MCP) over stdio using JSON-RPC 2.0 (protocol version 2024-11-05).

License

MIT