| name | notebooklm-auditor |
| description | Audit and optimize NotebookLM notebooks based on specific requirements. Use this skill when you need to verify if a notebook (sources, instructions, organization) matches a desired purpose, improve its retrieval quality, or prune redundant/noisy information. |
| license | Apache-2.0 |
| metadata | {"version":"1.0","capabilities":["auditing","optimization","notebooklm","iterative-refinement"]} |
NotebookLM Auditor Skill
This skill enables the agent to audit, optimize, and iterate on NotebookLM notebooks to ensure they meet specific domain requirements and quality standards.
Operational Flow
Follow this five-phase cycle for each iteration:
Phase 1: Reconnaissance
- List Sources: Use
notebooklm.notebook_get or source_list to identify all current sources.
- Identify Purpose: Read the notebook's personal instructions (via
notebook_get) and discuss with the user to refine the intended goal.
- Load Profile: Map the notebook to one of the profiles in
references/domain-profiles/.
Phase 2: Diagnosis
- Structural Analysis: Verify source completeness, formats, and sizes.
- Redundancy Detection: Identify sources with overlapping content (similarity > 0.85).
- Noise Identification: Flag sources irrelevant to the declared purpose.
- Terminology Check: Look for inconsistent naming or contradictions.
- Freshness Verify: Ensure dated content is current.
Phase 3: Planning
- Generate a prioritized plan:
- Remove redundant/noisy sources.
- Consolidate fragmented information.
- Add missing sources to cover gaps.
- Reorganize based on the domain's retrieval pattern.
- Update custom instructions (
configure_chat).
Phase 4: Execution with Checkpoint
- Create Checkpoint: Record the current source list and a baseline query response.
- Execute Changes: Apply the plan using tools like
source_delete, source_add, and configure_chat.
- Validate Result: If quality degrades, rollback by restoring the checkpoint state.
Phase 5: Validation & Convergence
- Stress Test: Run domain-specific queries to evaluate retrieval quality.
- Measure Metrics: Calculate scores for Context Relevance, Coverage, and Faithfulness.
- Check Convergence: Stop if metrics meet thresholds, max iterations (default 5) are reached, or plateau is detected.
Domain Heuristics
- Technical: Focused on symptoms, troubleshooting, and procedural accuracy.
- Business: Focused on market insights, JTBD, and competitive differentiation.
Principles & Limits
- Pruning: Less is often more. Prefer a clean, high-signal notebook over an exhaustive but noisy one.
- Safety: Never delete a source permanently without a checkpoint or user confirmation.
- Tool Limits: Be mindful of daily query limits in NotebookLM tiers.