Browser Automation

Enables AI agents to drive AutoCAD automation through one typed contract over two engines — live COM on Windows or headless ezdxf anywhere — to draft, edit and read drawings, including mechanical parts and sheets, architectural plans with measured room areas, and P&ID diagrams. It also critiques existing drawings, checks scales and topology, and produces takeoff workbooks.

Enables MCP agents to collaboratively direct a stick-figure flipbook animation through a single flipbook_step tool, handling story plans, motion plans, frame batches, and quality reviews as the animation advances in the browser.

Enables AI agents to manage Kuaituantuan (Pinduoduo community group buying) shops by publishing/delisting products and updating prices/stock, using official POP API with browser automation fallback.

Enables AI agents to search a catalog of MCP servers for physical hardware such as ESP32, Arduino, Raspberry Pi, 3D printers, Home Assistant devices, cameras, Zigbee and lab gear, returning install commands and ready-to-paste client configs. It also supports browsing device families and getting ranked recommendations for a specific device or model.

Enables coding agents to verify application changes by running browser workflows, executing independent state assertions, and producing replayable evidence.

Enables controlling QGIS to generate thematic maps over a project polygon, automatically styling and clipping layers, creating print layouts, and exporting them as PDF, PNG, JPG, or TIF. It also supports headless operation and map series with one sheet per layer.

Enables any MCP client or OpenAI-compatible application to use Meta Muse through a logged-in Chrome session, with tools for chat, attachments, session management, media retrieval, streaming, and tool-call compatibility.

Enables users to mint reusable local MCP integrations from website tasks and run the generated business tools through Codex, with chat-based job progress, input prompts, cancellation, and authenticated browser operations.

Enables AI agents to read a Proton Mail mailbox through a local Proton Bridge instance, listing folders, browsing recent messages, running full-text searches, and reading individual messages with decoded bodies and attachment info. All access is read-only — mailboxes are opened in IMAP READONLY mode, so nothing is ever modified, and the capability model lets operators opt into future draft, organize, send, and delete tools behind explicit confirmations.

Web content extraction API for AI agents. Scrape any URL and get clean, structured Markdown content with navigation, ads, and scripts stripped. Full JavaScript rendering via headless Chromium. Single and batch (10 URLs) modes. Built for RAG pipelines and AI research. Tools: web_scrape_to_markdown (single), web_scrape_batch (up to 10 URLs). Use this for RAG ingestion, research, content analysis, data extraction, or competitive intelligence. IMPORTANT: For screenshots/PDFs of pages, use capture_...

Moves attachments to and from the local disk without their bytes passing through the model's context, and manages several mail accounts under one owner so every write names its account explicitly. Enforces a server-side allow/ask/deny policy where risky writes create a pending action a person must approve in the browser, with no tool able to permanently delete mail.

Web capture API for AI agents. Take full-page screenshots (PNG/JPEG/WebP) and generate PDFs from any URL using headless Chromium. Custom viewport size, device emulation, and scroll capture. Tools: capture_screenshot (image), webpage_to_pdf (PDF). Use this for visual regression testing, documentation, archiving web pages, or generating reports. IMPORTANT: For text extraction from pages, use web_scrape_to_markdown. For custom PDF generation from data, use document_generate_pdf. Returns: base64-...

Enables structured access to YouTube via Piped, offering search, video metadata, and browser-ready playback URLs without an official API key or browser automation.

Enables headless, commandlet-based control of Ascent's Unreal project through small named tools for running scripts, probing and editing actors/assets, managing maps, materials, Ultra Dynamic Sky presets, importing assets, taking screenshots, and verifying workflows with safeguards.

Serves a local dsh workspace as a protected Streamable HTTP MCP endpoint, letting authenticated web AI agents read, search, patch, inspect, and operate the project, with optional shell commands, LSP support, and public tunneling.

Enables LLM agents to play a headless, deterministic 4X civilization game through ten MCP tools for turn briefs, unit orders, city management, research, and map exploration, with scoring against engine AI and do-nothing baselines.

Connects a local multimodal Ollama model to browser automation over MCP, letting an agent observe a page's DOM and screenshot, then plan and execute one safe Playwright action at a time until the task finishes. Includes a safety gate and audit log that halt irreversible actions for human handling.

Enables the Pi coding agent to drive Chrome through DevTools, providing browser automation tools, persistent or isolated profiles, screenshots, page analysis, and bundled agent policies.

Enables AI coding agents to control and inspect a live Chrome browser via MCP, including performance tracing, network and console debugging, screenshots, and reliable Puppeteer-based automation.

Enables AI agents to control, read, and verify a live Chrome browser through 46 schema-validated MCP tools, including DOM pruning, coordinate-aware interaction, and trusted CDP input events.

Enables Claude to read Canvas LMS deadlines, announcements, and grades using an existing browser session, with no admin token required.

Enables AI agents to drive a headless browser to a Storybook story and capture every custom event a component dispatches with zero registration, while flagging shared, retargeted, undocumented, and not-composed events. It also lets agents dispatch synthetic events at the rendered component to verify it responds as documented.

Enables searching restaurants on Tabelog and retrieving details, reviews, menus, ratings, photos, and online-booking availability as MCP tools, without a browser.

Enables LLM agents to read and navigate web pages efficiently with token budgets, extract structured data, and optionally control a browser with explicit user consent, remaining read-only by default.