NotebookLM MCP Server (Security Hardened)

# Changelog All notable changes to this project will be documented in this file. The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). ## [2026.2.3] - 2026-02-20 ### Fixed — Studio Panel Tools Fully Restored - **`generate_data_table` and `generate_video_overview`** now work correctly end-to-end, confirmed on macOS M4 (French locale, headless mode) - **Dead tile selector**: `clickDataTableTile` used `.mat-icon, [class*='icon']` which matched `SPAN.icon-container` before `<mat-icon>`, so `=== "table_view"` always failed silently. Fixed to `mat-icon` element tag (textContent is exactly `"table_view"`) - **False failure on slow shimmer**: after clicking the tile, if `shimmer-blue` didn't appear within 15s the tools returned `success: false`. Generation was triggering server-side but headless DOM update lagged. Now returns `{ success: true, status: "generating" }` so callers can poll - **`data-create-button-type` removed by Google** (Feb 2026): replaced with `mat-icon` text check and `jslog` numeric ID (`282298`) as locale-independent fallback - **Studio panel timeout** increased from 10s to 30s for slower machines and larger notebooks - **Full i18n pass**: all browser automation uses locale-independent signals first (CSS classes, Material icon names, element structure, `jslog` IDs) with English text as last-resort fallback only ### Added — CI / Branch Protection - GitHub Actions CI (`.github/workflows/ci.yml`) runs TypeScript build on every PR and push to `main` - `main` branch protection: force pushes blocked, branch deletion blocked, `Build` check required before merge ## [2026.2.2] - 2026-02-19 ### Fixed — Studio Panel Reliability on Slower Machines - **`generate_data_table` and `generate_video_overview`** no longer fail with "Could not find Studio panel toggle button" when the Studio panel loads collapsed or the DOM hasn't fully rendered - Added `waitForSelector` before Studio panel checks — blocks until either the panel tiles or toggle button appear (up to 10s), preventing race conditions on slower machines - Added `waitForSelector` for the generating artifact shimmer state after tile click — replaces fixed 3-4s delay, so generation is confirmed reliably regardless of machine speed - Multi-selector fallback chain retained for future-proofing against NotebookLM DOM changes ## [2026.2.1] - 2026-02-18 ### Fixed — Standard Profile Missing Key Tools - **Standard profile expanded** from 14 to 33 tools — all browser-based features now visible by default - Previously hidden tools now in standard: `create_notebook`, `batch_create_notebooks`, `add_source`, `remove_source`, `list_sources`, `generate_audio_overview`, `get_audio_status`, `download_audio`, `sync_library`, `remove_notebook`, `get_notebook_chat_history`, `get_query_history`, `re_auth`, `close_session`, `reset_session`, `get_quota`, `cleanup_data` - **Root cause**: The `standard` profile was never updated as new features were added, so key advertised features (notebook creation, source management, audio) were only available with `NOTEBOOKLM_PROFILE=full` - Gemini API tools remain in `full` profile only — keeps standard aligned with the "no API key required" promise - Full profile (`NOTEBOOKLM_PROFILE=full`) still includes all 47 tools (adds Gemini API, webhooks, compliance, export) ## [2026.2.0] - 2026-02-17 ### Added — Gemini 3 Model Support - **Gemini 3 models** — `gemini-3-flash-preview` and `gemini-3-pro-preview` now available as default models - **Deprecation warnings** — Using `gemini-2.5-flash` or `gemini-2.5-pro` now returns a warning that these models retire March 31, 2026 - **Incomplete status handling** — Deep Research now handles `"incomplete"` status from the API as a terminal state with partial results ### Added — Thinking Level Control - **`thinking_level` parameter** — New optional parameter for `gemini_query` and `deep_research` tools - Supports `minimal`, `low`, `medium`, and `high` levels for controlling response thoroughness vs speed ### Added — Structured JSON Output - **`response_schema` parameter** — New optional parameter for `gemini_query` tool - Pass a JSON schema to get structured, validated JSON responses from Gemini 3 - Automatically sets `responseMimeType: "application/json"` when schema is provided ### Added — Video Overview Generation - **`generate_video_overview`** — Generate AI-powered Video Overviews through NotebookLM's Studio panel - **`get_video_status`** — Check Video Overview generation progress - 10 visual styles: auto-select, custom, classic, whiteboard, kawaii, anime, watercolour, retro-print, heritage, paper-craft - 2 formats: explainer (full, 5-15 min) and brief (summary, 1-3 min) ### Added — Data Table Extraction - **`generate_data_table`** — Generate structured Data Tables from notebook sources via Studio panel - **`get_data_table`** — Extract generated table data as structured JSON (headers + rows) ### Changed - **Default model** changed from `gemini-2.5-flash` to `gemini-3-flash-preview` - **@google/genai SDK** upgraded from 1.38.0 to 1.41.0 - **Server banner** updated to reflect Gemini 3 ## [2026.1.12] - 2026-02-15 ### Security — Code Review & Medusa Scan Remediation - **Constant-time auth token comparison** using `secureCompare` (prevents timing attacks) - **Command injection fix** in `file-permissions.ts` — replaced `execSync()` with `execFileSync()` (array args) - **MCP SDK updated** to 1.26.0 — patches HIGH severity cross-client data leak (GHSA-345p-7cg4-v4c7) - **Audit hash chain** increased from 64-bit to 128-bit truncation for stronger collision resistance - **Settings JSON validation** — parsed settings now validated before merge (prevents property injection) - **Error message sanitization** — internal identifiers removed from error responses - **Dockerfile hardened** with `--no-install-recommends` - **Config env var validation** — `NOTEBOOK_PROFILE_STRATEGY` validated against allowed values ### Fixed — Memory Leaks & Concurrency - **CONFIG mutation race condition eliminated** — removed all 6 `Object.assign(CONFIG, ...)` call sites that could corrupt global state during concurrent requests - **RateLimiter memory leak** — empty keys now evicted from Map to prevent unbounded growth - **FinalizationRegistry self-reference** — fixed held value that prevented GC of secure buffers - **Event listener leak** — `framenavigated` listener now cleaned up after 30s timeout - **SecureCredential timer** — `.unref()` added so auto-wipe timer doesn't prevent process exit ### Performance - **Regex precompilation** in `sanitizeForLogging` — 5 patterns + email regex moved to module scope - **Response validator** — eliminated regex recompilation in `detectSuspiciousUrls` loop - **Rate limit detection** — consolidated 8+ IPC round-trips into single `page.evaluate()` call - **ESM import fix** — removed inline `require('path')` in favor of module-level import - **O(n) dedup** — notebook extraction uses Set-based deduplication instead of O(n^2) ### Code Quality - **Version strings unified** — MCP server and audit log now use `package.json` version - **Debounced library save** — `incrementUseCount` no longer writes to disk on every query - **Data URI pattern** — tightened false-positive-prone `data:` regex in response validator - **Quota storage** — moved from `configDir` to `dataDir` for consistency with directory lifecycle ## [2026.1.11] - 2026-02-02 ### Fixed - Notebook Sync Extraction for New Angular UI - **sync_library** now correctly extracts notebook UUIDs from NotebookLM's Angular Material UI - Automatically switches to grid view where notebook UUIDs are available in DOM element IDs - Primary strategy: extract from `project-button` card elements in grid view - Fallback: click-navigation through table rows to capture URLs - Last resort: basic table row extraction with placeholder URLs - **quota_manager** updated to detect notebooks via `project-button` (grid) and `project-action-button` (table) - Resolves issue reported in PR #3 — thanks @robert-merrill for identifying the UI change ### Added - Disable Gemini Tools Environment Variable - **NOTEBOOKLM_NO_GEMINI** - New environment variable to disable all Gemini API tools - Set `NOTEBOOKLM_NO_GEMINI=true` to hide 8 Gemini tools from tool list - Useful for clients with context window limitations (e.g., OpenCode) - Reduces tool count for clients that don't need Gemini features - Disabled tools: `deep_research`, `gemini_query`, `get_research_status`, `upload_document`, `query_document`, `list_documents`, `delete_document`, `query_chunked_document` ## [2026.1.10] - 2026-01-28 ### Fixed - Tool Description Clarity for Multi-LLM Compatibility - **ask_question** - Removed "Gemini" references that confused LLMs into thinking API key was needed - Now clearly states "Browser-Based • NO API KEY REQUIRED" - Added "PREFER THIS TOOL" guidance for notebook queries - **deep_research** - Added prominent warning "⚠️ REQUIRES GEMINI_API_KEY" - Added "When NOT to Use" section directing to ask_question - **gemini_query** - Added prominent warning "⚠️ REQUIRES GEMINI_API_KEY" - Added "When NOT to Use" section directing to ask_question - **upload_document** - Added note about alternatives that don't need API key This fix addresses feedback from OpenCode users where the LLM was incorrectly choosing Gemini API tools over browser-based tools. ## [2026.1.9] - 2026-01-28 ### Changed - Documentation & UX Improvements - **TL;DR Feature Summary** - Quick bullet list at top of README for instant understanding - **Updated "What's New in 2026"** - Shows all recent releases at a glance - **Full Feature List** - Collapsible section listing all 43 tools by category - **Gemini API Optional Callout** - Prominent messaging that core features need no API key - **Architecture Diagram** - Updated to show "NO API KEY NEEDED" vs "OPTIONAL" ### Security - Fixed 1 moderate vulnerability in hono dependency via `npm audit fix` ## [2026.1.8] - 2026-01-27 ### Changed - Major Dependency Updates - **@noble/post-quantum** 0.2.1 → 0.5.4 (FIPS 203/204/205 post-quantum cryptography updates) - **dotenv** 16.6.1 → 17.2.3 - **env-paths** 3.0.0 → 4.0.0 - **globby** 14.1.0 → 16.1.0 - **zod** 3.25.76 → 4.3.6 - **@types/node** 20.19.21 → 20.19.30 ### Fixed - **@noble/post-quantum import path** - Updated import from `@noble/post-quantum/ml-kem` to `@noble/post-quantum/ml-kem.js` (API change in v0.5.4) ## [2026.1.7] - 2026-01-27 ### Added - MCP Protocol UX Enhancements - **Tool Icons** - All 43 tools now have SVG icons for visual identification in compatible MCP clients - **Human-Friendly Titles** - Tools have proper display titles (e.g., "Ask NotebookLM" instead of "ask_question") - **Tool Behavior Annotations** - Tools include hints for client decision-making: - `readOnlyHint` - Indicates if tool only reads data - `destructiveHint` - Warns about data deletion operations - `idempotentHint` - Indicates if repeated calls are safe - `openWorldHint` - Shows if tool interacts with external services - **Task Support for Deep Research** - `deep_research` tool now declares `execution.taskSupport: "optional"` for proper long-running operation handling - **Resource Icons & Annotations** - Resources now include: - SVG icons for visual identification - `title` field for human-friendly display - `annotations` with `audience`, `priority`, and `lastModified` hints - **Predefined Prompts** - New prompts available via `prompts/list`: - `notebooklm.auth-setup` - Initial authentication guide - `notebooklm.auth-repair` - Authentication troubleshooting - `notebooklm.quick-start` - Getting started guide - `notebooklm.security-overview` - Security features documentation ### Changed - Updated `@modelcontextprotocol/sdk` from 1.25.2 to 1.25.3 - Updated `@google/genai` from 1.36.0 to 1.38.0 - Updated `patchright` from 1.55.0 to 1.57.0 - Updated `tsx` from 4.19.0 to 4.21.0 ### Security - Fixed 3 npm audit vulnerabilities (body-parser, hono, qs) ## [2026.1.4] - 2026-01-23 ### Security - **Defense-in-Depth Path Validation** - Added input validation for Windows `icacls` command - `isPathSafeForShell()` - Blocks shell metacharacters (`;&|`$` etc.) and path traversal (`..`) - `isUsernameSafe()` - Validates username format before shell use - Path normalization before execution - Addresses Medusa security scan finding (false positive but hardened anyway) ### Notes - Medusa scan showed 11 findings, 10 were false positives - This release hardens the one legitimate concern even though it wasn't exploitable ## [2026.1.3] - 2026-01-15 ### Changed - Updated `@modelcontextprotocol/sdk` from 1.0.0 to 1.25.2 ## [2026.1.2] - 2026-01-15 ### Added - Multi-Session Authentication Coordination - **Auth Lock System** - Global `.auth-in-progress` lock prevents race conditions when multiple Claude Code sessions authenticate simultaneously - **Wait for Auth** - Isolated profiles now wait for any in-progress authentication before cloning base profile - **Automatic State Reuse** - If another session completes auth while waiting, shared state is automatically reused ### Configuration New environment variables for multi-session support: ```bash export NOTEBOOK_PROFILE_STRATEGY=isolated # isolated|single|auto export NOTEBOOK_CLONE_PROFILE=true # Clone auth from base profile ``` ### How It Works 1. Session A starts auth → acquires lock → clears old auth → opens browser 2. Session B starts → needs isolated profile → detects lock → waits 3. Session A completes login → saves state → releases lock 4. Session B continues → clones now-authenticated profile → works immediately ## [2026.1.1] - 2026-01-14 ### Added - **Deep Health Check** - `get_health` tool now supports `deep_check: true` parameter to verify NotebookLM chat UI actually loads - Catches stale sessions where cookies exist but UI won't load ## [2026.1.0] - 2026-01-13 ### Added - **Chat History Context Management** - New `get_notebook_chat_history` tool for extracting conversation history from NotebookLM notebooks - **CalVer Versioning** - Switched to Calendar Versioning (2026.MINOR.PATCH) - Preview mode, pagination, and file export options for chat history ## [1.6.0] - 2025-12-18 ### Added - Enterprise Compliance Module Major release adding comprehensive enterprise compliance support for GDPR, SOC2 Type II, and CSSF (Luxembourg) regulations. #### Core Compliance Infrastructure (Phase 1) - **Compliance Logger** - Hash-chained audit logs with SHA-256 integrity verification - Tamper-evident logging with cryptographic chain - 7-year retention support (CSSF requirement) - Structured compliance events with actor tracking - **Data Classifier** - Automatic data sensitivity classification - 5 classification levels: PUBLIC, INTERNAL, CONFIDENTIAL, RESTRICTED, REGULATED - Pattern-based detection for PII, credentials, financial data - **Data Inventory** - GDPR Article 30 Records of Processing Activities - Automatic discovery and cataloging of all data stores - Processing purpose and legal basis tracking - **Consent Manager** - User consent tracking and management - GDPR Article 6 legal basis support - Consent versioning and expiration handling #### Data Subject Rights - GDPR (Phase 2) - **DSAR Handler** - Data Subject Access Request processing (Article 15) - Automated data collection and response generation - 30-day deadline tracking - **Data Erasure Manager** - Right to be forgotten (Article 17) - Verified secure deletion with audit trail - Scope-based erasure (categories, date ranges) - **Data Exporter** - Data portability (Article 20) - Machine-readable JSON export format - Checksum verification for integrity - **Retention Engine** - Automatic data retention enforcement - Configurable policies per data type - CSSF 7-year retention for audit logs #### Security Monitoring & Incident Response (Phase 3) - **Incident Manager** - Security incident lifecycle management - Severity-based workflow (low/medium/high/critical) - 72-hour notification deadline tracking (GDPR breach notification) - Root cause analysis and remediation tracking - **Alert Manager** - Multi-channel security alerting - Console, file, webhook, and email channels - Severity-based routing and rate limiting - **Breach Detector** - Pattern-based breach detection - Configurable detection rules - Automatic incident creation and alerting - **Health Monitor** - System availability monitoring (SOC2) - Component health checks - Uptime tracking and SLA reporting - **SIEM Exporter** - Enterprise SIEM integration - CEF (ArcSight), LEEF (QRadar), Syslog, Splunk HEC formats - Real-time event streaming #### Compliance Reporting & Documentation (Phase 4) - **Report Generator** - Compliance report generation - 10 report types: compliance_summary, gdpr_audit, soc2_audit, cssf_audit, security_audit, incident_report, dsar_report, retention_report, change_management, full_audit - JSON, CSV, HTML output formats - **Evidence Collector** - Audit evidence packages - Verifiable evidence with SHA-256 checksums - Regulation-specific collection (GDPR, SOC2, CSSF) - **Compliance Dashboard** - Real-time compliance status - Per-regulation status (compliant/at_risk/non_compliant) - Compliance score calculation (0-100) - CLI-formatted dashboard output - **Change Log** - Configuration change tracking (SOC2) - Before/after value tracking - Impact assessment (low/medium/high/critical) - Approval workflow support - **Policy Doc Manager** - Policy documentation management - 6 policy types: privacy, retention, access control, encryption, incident response, acceptable use - Version control and review scheduling - **16 MCP Compliance Tools** - Claude integration - Full compliance functionality exposed via MCP tools - Real-time compliance status queries ### Technical Details - **23 new TypeScript modules** in `src/compliance/` - **13,147 lines of code** for compliance functionality - **All modules use singleton pattern** for consistent state management - **Full type safety** with comprehensive TypeScript interfaces - **Zero external dependencies** for compliance code ### Documentation - Added `docs/COMPLIANCE-SPEC.md` - Full 4-phase implementation specification - Added MEDUSA scan response documenting false positive analysis ## [1.2.0] - 2025-11-21 ### Added - **Tool Profiles System** - Reduce token usage by loading only the tools you need - Three profiles: `minimal` (5 tools), `standard` (10 tools), `full` (16 tools) - Persistent configuration via `~/.config/notebooklm-mcp/settings.json` - Environment variable overrides: `NOTEBOOKLM_PROFILE`, `NOTEBOOKLM_DISABLED_TOOLS` - **CLI Configuration Commands** - Easy profile management without editing files - `npx notebooklm-mcp config get` - Show current configuration - `npx notebooklm-mcp config set profile <name>` - Set profile (minimal/standard/full) - `npx notebooklm-mcp config set disabled-tools <list>` - Disable specific tools - `npx notebooklm-mcp config reset` - Reset to defaults ### Changed - **Modularized Codebase** - Improved maintainability and code organization - Split monolithic `src/tools/index.ts` into `definitions.ts` and `handlers.ts` - Extracted resource handling into dedicated `ResourceHandlers` class - Cleaner separation of concerns throughout the codebase ### Fixed - **LibreChat Compatibility** - Fixed "Server does not support completions" error - Added `prompts: {}` and `logging: {}` to server capabilities - Resolves GitHub Issue #3 for LibreChat integration - **Thinking Message Detection** - Fixed incomplete answers showing placeholder text - Now waits for `div.thinking-message` element to disappear before reading answer - Removed unreliable text-based placeholder detection (`PLACEHOLDER_SNIPPETS`) - Answers like "Reviewing the content..." or "Looking for answers..." no longer returned prematurely - Works reliably across all languages and NotebookLM UI changes ## [1.1.2] - 2025-10-19 ### Changed - **README Documentation** - Added Claude Code Skill reference - New badge linking to [notebooklm-skill](https://github.com/PleasePrompto/notebooklm-skill) repository - Added prominent callout section explaining Claude Code Skill availability - Clarified differences between MCP server and Skill implementations - Added navigation link to Skill repository in top menu - Both implementations use the same browser automation technology ## [1.1.1] - 2025-10-18 ### Fixed - **Binary executable permissions** - Fixed "Permission denied" error when running via npx - Added `postbuild` script that automatically runs `chmod +x dist/index.js` - Ensures binary has executable permissions after compilation - Fixes installation issue where users couldn't run the MCP server ### Repository - **Added package-lock.json** - Committed lockfile to repository for reproducible builds - Ensures consistent dependency versions across all environments - Improves contributor experience with identical development setup - Enables `npm ci` for faster, reliable installations in CI/CD - Follows npm best practices for library development (2025) ## [1.1.0] - 2025-10-18 ### Added - **Deep Cleanup Tool** - Comprehensive system cleanup for fresh NotebookLM MCP installations - Scans entire system for ALL NotebookLM files (installation data, caches, logs, temp files) - Finds hidden files in NPM cache, Claude CLI logs, editor logs, system trash, temp backups - Shows categorized preview before deletion with exact file list and sizes - Safe by design: Always requires explicit confirmation after preview - Cross-platform support: Linux, Windows, macOS - Enhanced legacy path detection for old config.json files - New dependency: globby@^14.0.0 for advanced file pattern matching - CHANGELOG.md for version tracking - Changelog badge and link in README.md ### Changed - **Configuration System Simplified** - No config files needed anymore! - `config.json` completely removed - works out of the box with sensible defaults - Settings passed as tool parameters (`browser_options`) or environment variables - Claude can now control ALL browser settings via tool parameters - `saveUserConfig()` and `loadUserConfig()` functions removed - **Unified Data Paths** - Consolidated from `notebooklm-mcp-nodejs` to `notebooklm-mcp` - Linux: `~/.local/share/notebooklm-mcp/` (was: `notebooklm-mcp-nodejs`) - macOS: `~/Library/Application Support/notebooklm-mcp/` - Windows: `%LOCALAPPDATA%\notebooklm-mcp\` - Old paths automatically detected by cleanup tool - **Advanced Browser Options** - New `browser_options` parameter for browser-based tools - Control visibility, typing speed, stealth mode, timeouts, viewport size - Stealth settings: Random delays, human typing, mouse movements - Typing speed: Configurable WPM range (default: 160-240 WPM) - Delays: Configurable min/max delays (default: 100-400ms) - Viewport: Configurable size (default: 1024x768, changed from 1920x1080) - All settings optional with sensible defaults - **Default Viewport Size** - Changed from 1920x1080 to 1024x768 - More reasonable default for most use cases - Can be overridden via `browser_options.viewport` parameter - Config directory (`~/.config/notebooklm-mcp/`) no longer created (not needed) - Improved logging for sessionStorage (NotebookLM does not use sessionStorage) - README.md updated to reflect config-less architecture ### Fixed - **Critical: envPaths() default suffix bug** - `env-paths` library appends `-nodejs` suffix by default - All paths were incorrectly created with `-nodejs` suffix - Fix: Explicitly pass `{suffix: ""}` to disable default behavior - Affects: `config.ts` and `cleanup-manager.ts` - Result: Correct paths now used (`notebooklm-mcp` instead of `notebooklm-mcp-nodejs`) - Enhanced cleanup tool to detect all legacy paths including manual installations - Added `getManualLegacyPaths()` method for comprehensive legacy file detection - Finds old config.json files across all platforms - Cross-platform legacy path detection (Linux XDG dirs, macOS Library, Windows AppData) - **Library Preservation Option** - cleanup_data can now preserve library.json - New parameter: `preserve_library` (default: false) - When true: Deletes everything (browser data, caches, logs) EXCEPT library.json - Perfect for clean reinstalls without losing notebook configurations - **Improved Auth Troubleshooting** - Better guidance for authentication issues - New `AuthenticationError` class with cleanup suggestions - Tool descriptions updated with troubleshooting workflows - `get_health` now returns `troubleshooting_tip` when not authenticated - Clear workflow: Close Chrome → cleanup_data(preserve_library=true) → setup_auth/re_auth - Critical warnings about closing Chrome instances before cleanup - **Critical: Browser visibility (show_browser) not working** - Fixed headless mode switching - **Root cause**: `overrideHeadless` parameter was not passed from `handleAskQuestion` to `SessionManager` - **Impact**: `show_browser=true` and `browser_options.show=true` were ignored, browser stayed headless - **Solution**: - `handleAskQuestion` now calculates and passes `overrideHeadless` parameter correctly - `SharedContextManager.getOrCreateContext()` checks for headless mode changes before reusing context - `needsHeadlessModeChange()` now checks CONFIG.headless when no override parameter provided - **Session behavior**: When browser mode changes (headless ↔ visible): - Existing session is automatically closed and recreated with same session ID - Browser context is recreated with new visibility mode - Chat history is reset (message_count returns to 0) - This is necessary because NotebookLM chat state is not persistent across browser restarts - **Files changed**: `src/tools/index.ts`, `src/session/shared-context-manager.ts` ### Removed - Empty postinstall scripts (cleaner codebase) - Deleted: `src/postinstall.ts`, `dist/postinstall.js`, type definitions - Removed: `postinstall` npm script from package.json - Follows DRY & KISS principles ## [1.0.5] - 2025-10-17 ### Changed - Documentation improvements - Updated README installation instructions ## [1.0.4] - 2025-10-17 ### Changed - Enhanced usage examples in documentation - Fixed formatting in usage guide ## [1.0.3] - 2025-10-16 ### Changed - Improved troubleshooting guide - Added common issues and solutions ## [1.0.2] - 2025-10-16 ### Fixed - Fixed typos in documentation - Clarified authentication flow ## [1.0.1] - 2025-10-16 ### Changed - Enhanced README with better examples - Added more detailed setup instructions ## [1.0.0] - 2025-10-16 ### Added - Initial release - NotebookLM integration via Model Context Protocol (MCP) - Session-based conversations with Gemini 2.5 - Source-grounded answers from notebook documents - Notebook library management system - Google authentication with persistent browser sessions - 16 MCP tools for comprehensive NotebookLM interaction - Support for Claude Code, Codex, Cursor, and other MCP clients - TypeScript implementation with full type safety - Playwright browser automation with stealth mode