notebooklm-project-memory

Maintain one curated NotebookLM notebook per project and export cited evidence packs.

Updated Jul 19, 2026
One-click install
npx skills add https://github.com/DagimAlemayehuu/AgenticEngineering --skill notebooklm-project-memory-dagimalemayehuu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: notebooklm-project-memory
Source: https://github.com/DagimAlemayehuu/AgenticEngineering/tree/main/agentic-engineering/core/dependencies/research/notebooklm-project-memory
Command: npx skills add https://github.com/DagimAlemayehuu/AgenticEngineering --skill notebooklm-project-memory-dagimalemayehuu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and assets (resource) components.

What problem does it solve? Research for software projects often scatters across ad-hoc notebooks, uncited web claims, and duplicated sources, making it hard to ground architecture and planning decisions in verifiable evidence. This Skill enforces exactly one curated NotebookLM notebook per project and turns bounded research into cited, local evidence packs that planning, Jules issues, and PR reviews can consume. ## Core Features & Use Cases - Single-notebook enforcement: Binds each project to one canonical NotebookLM notebook via an immutable local registry that refuses a second binding and validates source and evidence policies. - Curated source management: Organizes canonical, working, and evidence source classes with deduplication, redaction, a 60-source active limit, and a 20-source per-request import cap. - Research-to-delivery loop: Converts bounded research questions into a cited evidence pack with findings, recommendation, constraints, uncertainty, and provenance for plans and PR reviews. - Use Case: Before migrating a dependency, onboard the project to its canonical notebook, import up to 20 primary documentation sources, query for constraints, and hand the resulting evidence pack to the planning phase and the Jules issue. ## Quick Start Use $notebooklm-project-memory to bind this project to one NotebookLM notebook and prepare a cited evidence pack for the current decision.

Frequently Asked Questions about notebooklm-project-memory

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I set up one NotebookLM notebook for a software project?▼

Create the notebook with nlm notebook create, set an alias with nlm alias set, then run manage_project_notebook.py init to bind the project in a local registry. The registry refuses any different notebook ID afterward, enforcing one notebook per project.

How do I turn NotebookLM research into a cited evidence pack?▼

Fill the research request template with one decision, known facts, questions, and source boundaries, then query the notebook and save results using the evidence pack template. The pack must include citations, a recommendation, constraints, uncertainty, and provenance.

What sources should not be uploaded to a project NotebookLM notebook?▼

Forbidden sources include secrets, credentials, raw production data, unredacted customer data, and entire repository dumps. The registry validation rejects any source policy that weakens these forbidden source classes.

How many sources can a NotebookLM project notebook contain?▼

The policy caps the notebook at 60 active sources and limits any single research request to 20 imported sources. The validation script enforces that the per-request limit never exceeds the active source limit.

When should I not use NotebookLM for project research?▼

Avoid it for ordinary code navigation, unverified web claims, code changes, secrets, or full repository uploads. NotebookLM answers must never override checked local contracts, test results, or primary sources.