autonotebooks

name	AutoNotebooks
description	A semi-skill for parallelized NotebookLM notebook creation and artifact generation through AI-User interaction.

⚠️ Critical Rule: No Code Modifications

The automation engine (nlm_runner.py and utils/) is a finalized, robust system that supports ALL NotebookLM features through its configuration schema. As an AI agent, you MUST NOT modify any Python code in this directory.

Your sole responsibility is to:

1. Gather requirements.
2. Generate the correct JSON configuration.
3. Execute the pipeline using the --config flag.

---

Workflow: Back-and-Forth Planning

Phase 0: Ensure Requirements are Setup

You will need to ensure the user has two python libraries available: notebooklm-mcp and rich.

• If you are running in an environment where you can execute commands, run:

  uv sync
  

Before generating any files or running the automation, you must gather all necessary details from the user.

Phase 1: Requirement Gathering

Ask for the following details (grouping is allowed for efficiency):

1. Topics: Names and research queries. NEW: You can now also provide an existing notebook_id to skip creation, specific sources (URLs, local files, Drive IDs, or YouTube links), and per-topic artifacts to generate different artifacts for different notebooks.
2. Research: Mode (fast/deep), Source (web/drive), and if it should be force started.
3. Chat Settings: Optional goal (learning_guide, custom) and response_length.
4. Artifacts: Which ones? (audio, video, slides, report, quiz, flashcards, mindmap, infographic, data-table).
5. Artifact Specs:
   • Flags: (e.g., audio format, quiz difficulty).
   • Focus: Specific focus prompt for this artifact.
   • Sources: Filter which source IDs to use for this artifact.
   • Revision: (For slides) specific slide-by-slide revision instructions.
   • Rename: Custom title for the generated artifact.
6. Downloads: Whether to download locally (Default: yes).

Phase 2: Confirmation

Consolidate the plan and ask for user confirmation.

Phase 3: Execution

1. Create Config Directory: NotebookAutomation/config/.
2. Write Config: config_<timestamp>.json following the PipelineConfig schema in utils/config.py.
3. Run Pipeline: Locate nlm_runner.py (check root or .agent/rules/) and execute strictly using the --config flag:

   uv run python [path/to/]nlm_runner.py --config config/config_<timestamp>.json
   

Advanced Config Schema Reference

{
  "topics": [
    {
      "key": "unique_id",
      "title": "Notebook Title",
      "query": "Research Query",
      "sources": [{ "type": "url", "value": "https://..." }],
      "chat": { "goal": "learning_guide", "response_length": "longer" },
      "artifacts": [
        {
          "type": "audio",
          "flags": { "format": "debate" },
          "rename": "Deep Dive"
        },
        { "type": "infographic" }
      ]
    },
    {
      "key": "another_topic",
      "title": "Another Notebook",
      "query": "Another query",
      "artifacts": [{ "type": "audio" }]
    }
  ],
  "research_mode": "fast",
  "artifacts": [],
  "download": true
}

Artifact Resolution: If a topic defines its own artifacts list, it overrides the global artifacts for that topic. If a topic omits artifacts (or sets it to null), the global artifacts list is used. The terminal dashboard will show a neutral ─ for any artifact column that does not apply to a given topic.