"browser use" matching MCP servers:

A standalone MCP server for Linux desktop control, enabling screenshots, mouse/keyboard input, app/window management, and clipboard access via a Python bridge.

A self-hosted MCP server that enables AI agents to see and control a Windows PC through screenshots, built-in Arabic/English OCR, mouse, keyboard, clipboard, and application launching, with code-enforced consent and audit logging.

Enables agents to record their Linux X11 display while doing browser or computer-use work, then return a replay link when the task finishes without sending the video itself to the model.

Gives AI agents a compact, semantic interface to the browser, returning structured page snapshots with stable element IDs instead of raw DOM. Enables agents to navigate, interact, and extract information from web pages efficiently.

Enables web search via a real browser for any LLM, bypassing anti-bot measures without API keys.

Enables inspection and control of Arc Browser spaces and tabs on macOS, including reading sidebar/history data and performing actions like focusing spaces, creating and closing tabs.

Enables AI agents to fully control a browser for web automation, including navigation, clicking, typing, scrolling, screenshots, and DOM inspection, with session persistence and anti-bot bypass.

Exposes a headful, collaborative browser over MCP so an agent and a user can share one visible Chromium window: the agent navigates, reads, clicks, types, and fills forms, while the user watches and can take over for passwords, CAPTCHAs, 2FA, and payments, with every action written to an append-only audit trail. It also runs headless under a chat-based harness.

Lets an AI assistant inspect a running Google Chrome or Microsoft Edge browser over the Chrome DevTools Protocol, reading console output, network traffic, DOM snapshots, errors, performance metrics, storage, and screenshots. It also supports live event subscriptions, screen-frame streaming, and one-shot export of everything into an offline-readable archive such as HAR, CSV, and a self-contained HTML report.

An MCP server for interacting with the USAspending.gov API, supporting local use with Claude Desktop and cloud deployment with login.gov authentication.

Use your signed-in Chrome through a local MCP server and companion extension. Supports tabs, navigation, forms, screenshots, and page reading, with a deny-all domain allowlist. Install: npx -y mcp-browser-extension@latest.

Enables MCP clients to retrieve normalized U.S. public statistics — Census ACS population, BLS unemployment and CPI, FRED series, and HUD fair market rents — as small, typed JSON snapshots. Exposes five tools (get_cbsa_population, get_hud_fmr, get_unemployment_rate, get_cpi, get_fred_series) that work offline against recorded fixtures and can later use live API keys without changing tool names.

Standalone MCP server exposing only the security-classified tools from Shade Browser, enabling Brave CDP browser automation across 20 isolated profile slots. Built on FastMCP and Playwright Python.

parallel-browser-mcp is an MCP server for parallel browser automation. It exposes a numeric session model over MCP so one client can create and control multiple browser sessions at the same time across multiple browser providers.

Perfume database API with verified notes: 1,500+ perfumes, 160+ houses, 400+ raw materials in a fragrance ingredients database. 12 tools over MCP and REST, OpenAPI spec, free and keyless, answers in 10 languages. Look up a note or raw material and get the perfumes that use it, or a perfume and get its full top, heart and base pyramid. Built by Scento.

Google Hotels MCP server via direct RPC — no scraping, no browser automation. Three tools: hotel list search (16 filter slots: stars, price, amenities, brands, free cancellation), per-OTA rate plans and cancellation policies for a single hotel, and parallel top-N enrichment. One-command setup for Claude Code, Codex, and ChatGPT

An MCP front-end for AnB. The tool surface is "use-don't-reveal": no tool returns a plaintext secret. Reveal paths require a TTY the server doesn't have, so the no-reveal guarantee is structural, not prompt-based — it holds even against a prompt-injected agent. Runs as a dedicated, narrowly-scoped AnB identity; exec is default-deny (only allowlist rules tagged scope=mcp run); command output

Bridges the editing capabilities of an already-open DingTalk document in the browser to any MCP host via standard stdio, forwarding tools to the page's internal MCP server over a local WebSocket. Supports read/write operations through suggested edits, pairing-based security, and zero runtime dependencies.

Enables AI coding assistants to debug and analyze JavaScript in web pages by managing scripts, breakpoints, network and WebSocket traffic, browser state, and local file inputs/outputs for continuous reverse-engineering workflows.

Enables AI assistants to run code in 24 programming languages on HackerEarth's sandboxed servers and read the printed output or errors, without installing local toolchains. It also lists supported languages with exact identifiers for use when submitting code.