Capmonster Captcha Solver MCP server

CapMonster Cloud MCP Server (Model Context Protocol)

An Model Context Protocol (MCP) server for CapMonster Cloud, available as a Python package and as a TypeScript port (ts/).

This server is the solve brain: it lists supported captcha types, serves CapMonster's live docs, and creates/polls solve tasks against the CapMonster Cloud REST API. It has no browser of its own — pair it with a browser-driving MCP (e.g. capmonster-mcp-patchright) that does the page work (navigation, interaction, reading the live DOM/network, and injecting the solved token back into the page). See capmonster_agent/SKILL.md for the full step-by-step procedure for analyzing a captcha-protected page and solving it this way.

👉 Get your Free API Key and Start Bypassing CAPTCHAs

⚡ Supported CAPTCHAs

Your AI Agent will be able to automatically bypass, among others:

TenDI, Hunt, Altcha, Basilisk, and more

• reCAPTCHA (v2, v2 Enterprise, v3)
• Cloudflare Turnstile and Cloudflare Challenge (managed challenge / cf_clearance)
• FunCaptcha (Arkose)
• GeeTest (v3 and v4)
• Enterprise Anti-Bot Systems: AWS WAF, DataDome, Imperva, TSPD, Binance, Prosopo, Yidun,

• Image-to-Text & Complex Image Tasks

The authoritative, current list is served live from CapMonster's OpenAPI spec via the getsupportedtasks tool — hCaptcha is not currently supported, despite appearing in some of CapMonster's own marketing copy.

📦 Installation

Requires Python 3.11+ and a valid CapMonster API Key. Run it with uvx (no local clone needed):

uvx capmonster-mcp

Prefer TypeScript/Node? See ts/README.md for the npm-published port — same tools, same behavior.

🔌 Using with an MCP client (e.g. Claude Desktop)

Add the following to your MCP client's config (e.g. claudedesktopconfig.json):

{
  "mcpServers": {
    "capmonster": {
      "command": "uvx",
      "args": ["capmonster-mcp"],
      "env": {
        "CM_API_KEY": "your_api_key_here"
      }
    }
  }
}

The server only runs over stdio, so there are no HTTP headers to carry a per-request key — CMAPIKEY is read once from the environment and shared by every tool call in the session.

Pair it with a browser-automation MCP server (e.g. capmonster-mcp-patchright) so your agent can both see the page and solve what's on it — see mcp.example.json for a config with both servers wired up together.

🛠 Available MCP Tools

Once connected, your LLM will have access to the following tools:

schema for a task type.

• getsupportedtasks: Lists captcha task types CapMonster supports, from the live OpenAPI spec.
• gettaskparameters(task_type): Required/optional fields, variant notes, and the solution

(docs.capmonster.cloud / api.capmonster.cloud only), with section-jump and pagination.

• get_docs(url, offset, limit, section): Fetches a CapMonster documentation page

completion (preferred over driving the poll loop yourself).

• create_task(task): Submits a captcha task and returns a taskId.
• gettaskresult(task_id): Polls a task once.
• gettaskresultwait(taskid, timeoutseconds, pollinterval_seconds): Polls a task to

fingerprint across the browser and the solve task.

• getactualuser_agent(): Fetches a current Windows User-Agent to use as one consistent

• get_balance(): Checks your current CapMonster Cloud API balance.

📚 Official Documentation

Shortened. The full README is on GitHub.