NotebookLM MCP Management
Tech Stack Target / Version: NotebookLM current web release, Markdown session capture, and URL-tracked research workflows.
Use this skill when research should be grounded in NotebookLM notebooks instead of a generic web search.
- Leverage native parallel subagent dispatch and 200k+ context windows where available.
Current MCP Reality
This repository already targets a real NotebookLM MCP workflow. The concrete tool surface available in this environment includes:
get_health
list_notebooks, search_notebooks, select_notebook
ask_question
add_notebook, update_notebook, remove_notebook
list_sessions, reset_session, close_session
setup_auth, re_auth, cleanup_data
Activation Conditions
Use symptom -> action triggers: when one matches, apply this skill and verify with the protocol below.
- Querying a specific NotebookLM notebook
- Adding a notebook from a share URL
- Managing a notebook library or switching active notebooks
- Recovering authentication or cleaning NotebookLM state
- Continuing a multi-turn research session
Recommended Workflow
- Call
get_health first to confirm authentication and server readiness.
- Reuse an existing session when the task is the same.
- Prefer
search_notebooks or list_notebooks before asking the user to restate what is already in the library.
- Use
ask_question iteratively in the same session for deep work.
- Use
setup_auth or re_auth only when health indicates auth problems.
Library Management Rules
- Do not add or remove notebooks without explicit user confirmation.
- When adding a notebook, collect URL, description, topics, and use cases first.
- Update metadata instead of creating duplicates when the notebook already exists.
Troubleshooting
- Auth broken:
get_health -> re_auth
- Stale browser state:
cleanup_data(preserve_library=true) after closing browsers
- Wrong context:
reset_session or switch notebooks
- Ambiguous notebook choice: search the library before creating a new one
Cross-Client Portability
This skill is written to stay usable across GitHub Copilot, Claude Code, and Codex.
- GitHub Copilot: keep the folder in a Copilot-visible skill path or wrap the
workflow in project instructions when folder discovery is unavailable.
- Claude Code: keep the folder in a local skills directory or a compatible plugin source.
- Codex: install or sync the folder into
$CODEX_HOME/skills/notebooklm-management and restart Codex after major changes.
MCP Availability And Fallback
Preferred MCP Server: NotebookLM MCP
- Fallback prompt: "Use the NotebookLM MCP Management skill without MCP. Follow the documented local or manual fallback, show the selected tool surface, and report the verification evidence."
- Use the NotebookLM web UI directly, capture answers in Markdown, and store session notes locally when the MCP server is unavailable.
- Preserve notebook URLs, prompt history, and manual research notes so the workflow remains reproducible.
- Do not claim an MCP operation was used when the active host does not expose it.
Anti-Patterns
- Treating source content as already clean: Formatting automation will happily preserve broken or inconsistent input.
- Skipping an open-file verification pass: Documents and spreadsheets often fail in the destination app, not in the script output.
- Automating irreversible edits without checkpoints: A small mapping mistake can affect an entire workbook or document.
Verification Protocol
Before claiming "skill applied successfully":
- Pass/fail: The Notebooklm Management workflow names the agent boundary, delegated scope, and expected return artifact.
- Pass/fail: Context passed to helpers is minimal, task-local, and free of hidden expected answers.
- Pass/fail: Results are integrated only after evidence, diffs, or citations are checked by the controller.
- Pressure-test scenario: Run the workflow on two similar tasks that must not share assumptions or leaked context.
- Success metric: Zero context leakage; every delegated output is independently reviewable.
References & Resources
Documentation
Scripts
Examples
Related Skills
- documentation-authoring: Use it when the workflow also needs drafting structured technical or product documents.
- notion-docs: Use it when the workflow also needs Notion page and database publishing workflows.
- pdf: Use it when the workflow also needs PDF extraction, generation, and layout-aware review.
- word-document: Use it when the workflow also needs Word document authoring and formatting workflows.
1---
2name: notebooklm-management
3version: "2.0"
4last_updated: 2026-09-08
5tags: [research, management, documents, automation, productivity]
6description: "NotebookLM MCP server management - query notebooks, add from share links, handle auth, reset sessions. Use when working with Google NotebookLM notebooks for conversational research tasks."
7metadata:
8 source: https://skillmd.com/skills/practicalswan/notebooklm-management
9---
10
11# NotebookLM MCP Management
12
13> Tech Stack Target / Version: NotebookLM current web release, Markdown session capture, and URL-tracked research workflows.
14
15Use this skill when research should be grounded in NotebookLM notebooks instead of a generic web search.
16
17- Leverage native parallel subagent dispatch and 200k+ context windows where available.
18
19
20## Current MCP Reality
21
22This repository already targets a real NotebookLM MCP workflow. The concrete tool surface available in this environment includes:
23
24- `get_health`
25- `list_notebooks`, `search_notebooks`, `select_notebook`
26- `ask_question`
27- `add_notebook`, `update_notebook`, `remove_notebook`
28- `list_sessions`, `reset_session`, `close_session`
29- `setup_auth`, `re_auth`, `cleanup_data`
30
31## Activation Conditions
32
33Use symptom -> action triggers: when one matches, apply this skill and verify with the protocol below.
34
35- Querying a specific NotebookLM notebook
36- Adding a notebook from a share URL
37- Managing a notebook library or switching active notebooks
38- Recovering authentication or cleaning NotebookLM state
39- Continuing a multi-turn research session
40
41## Recommended Workflow
42
431. Call `get_health` first to confirm authentication and server readiness.
442. Reuse an existing session when the task is the same.
453. Prefer `search_notebooks` or `list_notebooks` before asking the user to restate what is already in the library.
464. Use `ask_question` iteratively in the same session for deep work.
475. Use `setup_auth` or `re_auth` only when health indicates auth problems.
48
49## Library Management Rules
50
51- Do not add or remove notebooks without explicit user confirmation.
52- When adding a notebook, collect URL, description, topics, and use cases first.
53- Update metadata instead of creating duplicates when the notebook already exists.
54
55## Troubleshooting
56
57- Auth broken: `get_health` -> `re_auth`
58- Stale browser state: `cleanup_data(preserve_library=true)` after closing browsers
59- Wrong context: `reset_session` or switch notebooks
60- Ambiguous notebook choice: search the library before creating a new one
61
62<!-- MCP:START -->
63
64<!-- PORTABILITY:START -->
65## Cross-Client Portability
66
67This skill is written to stay usable across GitHub Copilot, Claude Code, and Codex.
68
69- GitHub Copilot: keep the folder in a Copilot-visible skill path or wrap the
70 workflow in project instructions when folder discovery is unavailable.
71- Claude Code: keep the folder in a local skills directory or a compatible plugin source.
72- Codex: install or sync the folder into
73 `$CODEX_HOME/skills/notebooklm-management` and restart Codex after major changes.
74
75<!-- PORTABILITY:END -->
76
77## MCP Availability And Fallback
78
79Preferred MCP Server: NotebookLM MCP
80
81- Fallback prompt: "Use the NotebookLM MCP Management skill without MCP. Follow the documented local or manual fallback, show the selected tool surface, and report the verification evidence."
82- Use the NotebookLM web UI directly, capture answers in Markdown, and store session notes locally when the MCP server is unavailable.
83- Preserve notebook URLs, prompt history, and manual research notes so the workflow remains reproducible.
84- Do not claim an MCP operation was used when the active host does not expose it.
85
86<!-- MCP:END -->
87
88## Anti-Patterns
89
90- Treating source content as already clean: Formatting automation will happily preserve broken or inconsistent input.
91- Skipping an open-file verification pass: Documents and spreadsheets often fail in the destination app, not in the script output.
92- Automating irreversible edits without checkpoints: A small mapping mistake can affect an entire workbook or document.
93
94## Verification Protocol
95
96Before claiming "skill applied successfully":
97
981. Pass/fail: The Notebooklm Management workflow names the agent boundary, delegated scope, and expected return artifact.
992. Pass/fail: Context passed to helpers is minimal, task-local, and free of hidden expected answers.
1003. Pass/fail: Results are integrated only after evidence, diffs, or citations are checked by the controller.
1014. Pressure-test scenario: Run the workflow on two similar tasks that must not share assumptions or leaked context.
1025. Success metric: Zero context leakage; every delegated output is independently reviewable.
103
104## References & Resources
105
106### Documentation
107- [MCP Tool Reference](./references/mcp-tool-reference.md) - Current NotebookLM MCP operations and parameters
108- [Troubleshooting Guide](./references/troubleshooting.md) - Auth recovery, cleanup, and session issues
109- [Workflows](./references/workflows.md) - Library, query, and maintenance workflows
110
111### Scripts
112- [NotebookLM Helper](./scripts/notebooklm-helper.py) - Local helper for library exports and reporting when MCP access is unavailable
113- [Scripts README](./scripts/README.md) - Quick commands for the helper script
114
115### Examples
116- [simple-query.py](./examples/simple-query.py) - Basic query pattern
117- [multi-turn-conversation.py](./examples/multi-turn-conversation.py) - Session reuse pattern
118- [library-management.py](./examples/library-management.py) - Library search and organization
119
120## Related Skills
121
122- [documentation-authoring](../documentation-authoring/SKILL.md): Use it when the workflow also needs drafting structured technical or product documents.
123- [notion-docs](../notion-docs/SKILL.md): Use it when the workflow also needs Notion page and database publishing workflows.
124- [pdf](../pdf/SKILL.md): Use it when the workflow also needs PDF extraction, generation, and layout-aware review.
125- [word-document](../word-document/SKILL.md): Use it when the workflow also needs Word document authoring and formatting workflows.