What problem does it solve? NotebookLM research operations fail unpredictably: MCP servers enter cooldowns, task IDs go stale after notebook deletion or consolidation, and parallel imports crash the server. This Skill provides a provider-neutral adapter with documented failure modes, CLI fallbacks, and evidence-preservation rules so research workflows stay recoverable. ## Core Features & Use Cases - Lifecycle Guardrails: Verifies host capability and approved notebook IDs before any operation, treats notebook outputs as untrusted evidence, and requires human approval for deletions. - Failure Recovery Playbooks: Reference notes document MCP cooldown cascades, full server unreachability with valid auth, stale task ID detection, and sequential import patterns with the nlm CLI fallback. - Maintenance Automation: The notebooklm_helper.py script deduplicates sources across notebooks by URL and title, and imports completed research into empty notebooks. - Use Case: A cron job monitoring 12 research notebooks hits MCP timeouts; the Skill's decision tree routes status checks to the nlm CLI with --max-wait 0, completing all checks in under a minute without triggering cooldown. ## Quick Start Use the notebooklm-management skill to check research status across my notebooks and import any completed sources, falling back to the nlm CLI if MCP is unreachable.
