NotebookLM MCP

Create Audio Overview

audio_overview_create

Generate a podcast-style conversation between two AI hosts from a NotebookLM notebook's sources. Confirm to start; it uses your quota and takes minutes.

Instructions

Start generating an Audio Overview: a podcast-style conversation between two AI hosts about the notebook's sources. Uses the user's NotebookLM quota and takes several minutes, so agree on the options with the user first: with confirm=false it does nothing and returns pending_confirmation. With confirm=true it returns the artifact_id; poll studio_status until that artifact is completed. Once completed, studio_status gives an audio download_url.

Input Schema

Name	Required	Description
`format`	No	Format: brief, critique, debate, deep_dive. deep_dive is the default two-host discussion
`length`	No	Length: default, long, short
`confirm`	Yes	Must be true to start generation. With false the tool only returns pending_confirmation and does nothing
`language`	No	BCP-47 language code such as en, es or fr; omit for NotebookLM's default
`source_ids`	No	Limit generation to these source IDs (from notebook_get); omit to use every source in the notebook
`notebook_id`	Yes	Notebook ID, as returned by notebook_list or notebook_create
`focus_prompt`	No	Optional instruction on what to focus on or emphasise, e.g. "the methodology section"

Output Schema

Name	Required	Description
`status`	Yes
`message`	No	Human-readable outcome or next step
`artifact_id`	No	ID of the new artifact, as listed by studio_status

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedv0.5.0Oct 20, 2026

changedOutput schema / (root)

Previous value: -nullNew value: +{
+  "$schema": "http://json-schema.org/draft-07/schema#",
+  "additionalProperties": false,
+  "properties": {
+    "artifact_id": {
+      "description": "ID of the new artifact, as listed by studio_status",
+      "type": "string"
+    },
+    "message": {
+      "description": "Human-readable outcome or next step",
+      "type": "string"
+    },
+    "status": {
+      "enum": [
+        "success",
+        "pending_confirmation"
+      ],
+      "type": "string"
+    }
+  },
+  "required": [
+    "status"
+  ],
+  "type": "object"
+}

Changed6 schema fields changedv0.4.0Oct 20, 2026

changedInput schema / properties / confirm / description

Previous value: -"Must be true to start generation"New value: +"Must be true to start generation. With false the tool only returns pending_confirmation and does nothing"

changedInput schema / properties / focus_prompt / description

Previous value: -"Focus prompt for the audio"New value: +"Optional instruction on what to focus on or emphasise, e.g. \"the methodology section\""

changedInput schema / properties / format / description

Previous value: -"Format: brief, critique, debate, deep_dive"New value: +"Format: brief, critique, debate, deep_dive. deep_dive is the default two-host discussion"

changedInput schema / properties / language / description

Previous value: -"BCP-47 language code (e.g. en, es)"New value: +"BCP-47 language code such as en, es or fr; omit for NotebookLM's default"

changedInput schema / properties / notebook_id / description

Previous value: -"The notebook ID"New value: +"Notebook ID, as returned by notebook_list or notebook_create"

changedInput schema / properties / source_ids / description

Previous value: -"Source IDs (omit for all)"New value: +"Limit generation to these source IDs (from notebook_get); omit to use every source in the notebook"

First observedv0.3.1Aug 6, 2026

TDQS

A4.7/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only say readOnly=false, openWorld=true, idempotent=false, destructive=false. The description adds what actually matters for a long-running job: it consumes user quota, runs for several minutes, is gated by a confirmation flag, is asynchronous, and the result must be polled via studio_status before a download_url exists. None of that is recoverable from the annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Three dense sentences, front-loaded with what the tool produces before the operational caveats. No filler, no repetition of the title, and every clause carries actionable information.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

An output schema exists so return values need no explanation, and the description still closes the loop on async completion (poll studio_status, then fetch download_url). For a 7-parameter, non-idempotent, quota-consuming generation tool, nothing an agent needs to call it correctly is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so every parameter including confirm, format, source_ids, and language is already documented in the schema. The description restates the confirm contract and the provenance of source_ids and notebook_id rather than adding new syntax or constraints. Baseline 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Start generating an Audio Overview') and immediately defines the artifact: 'a podcast-style conversation between two AI hosts about the notebook's sources'. That level of detail is enough to separate it from siblings like video_overview_create, infographic_create, or slide_deck_create without opening any schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives the preconditions ('uses the user's NotebookLM quota and takes several minutes, so agree on the options with the user first'), the exact two-phase invocation contract (confirm=false returns pending_confirmation, confirm=true returns artifact_id), and the follow-up path ('poll studio_status until that artifact is completed'). It also implicitly routes the agent to notebook_get for source_ids and studio_status for retrieval.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.