Capsolver Mcp

io.github.capsolver-aiPublished by GitHub account capsolver-aiv0.1.3Updated Oct 24, 2026

MCP Server for CapSolver - captcha-solving capabilities for AI agents

VerifiedSTDIODesktop onlyBrowser AutomationSecurity & Monitoring

Overview

AI-generated overview

A local MCP server that gives AI assistants CapSolver captcha-solving tools, including page detection and autofill, using a CapSolver API key.

What you can do
Solve captchas by type and site parameters in token mode, and list supported captcha types and handlers. Scan a page URL to detect which captcha types are present. Detect, solve and autofill captchas on a page using a Playwright browser. Check the CapSolver account balance and packages.
When to use it
Use it when an assistant needs to get past captchas during automated browsing or form workflows, or when you want to check which captcha types CapSolver supports and what your account balance is.
Requirements
A local Python runtime; the package is published on PyPI and can be run with uvx. A CapSolver account and API key supplied through the CAPSOLVER_API_KEY environment variable. The browser-based tools need the browser extra and a Playwright Chromium install. Network access to CapSolver.
Before you install
Requires the CAPSOLVER_API_KEY secret, which is sent to CapSolver. Solving captchas consumes paid CapSolver credits, so check the balance first. The page tools open pages and autofill forms, which changes page state. Solving captchas may violate the terms of service of the sites involved.

Installation

In SourceWeft

  1. Open Capsolver Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

capsolver-mcp

MCP Server for CapSolver — expose captcha-solving capabilities to AI agents via the Model Context Protocol.

Published on PyPI as capsolver-mcp and listed in the official MCP Registry as io.github.capsolver-ai/capsolver-mcp.

See the capsolver-ai-hub repo for integration examples and the full documentation.

For detailed MCP client setup (Claude Desktop, Claude Code, Cursor, Windsurf, Cline, and more), see docs/mcp-integration.md.

Install

bash
pip install capsolver-mcppip install capsolver-mcp[browser]   # with Playwright support (for detect/solve_on_page)

All tools read the API key from the environment:

bash
# bash / zshexport CAPSOLVER_API_KEY="your-capsolver-api-key"
# PowerShell$env:CAPSOLVER_API_KEY = "your-capsolver-api-key"
# cmdset CAPSOLVER_API_KEY=your-capsolver-api-key

Usage

CLI

bash
# stdio (default — for local MCP clients like Claude Desktop)capsolver-mcp
# SSE (for remote / HTTP access)capsolver-mcp --transport sse --host 0.0.0.0 --port 8000
# Streamable HTTP (MCP 2025-03-26 spec)capsolver-mcp --transport streamable-http --host 0.0.0.0 --port 8000
CLI options
capsolver-mcp [OPTIONS]
  --transport {stdio,sse,streamable-http}                              Transport protocol (default: stdio)  --host HOST                 Bind host for SSE/HTTP transports (default: 127.0.0.1)  --port PORT                 Bind port for SSE/HTTP transports (default: 8000)  --api-key KEY               API key (fallback: CAPSOLVER_API_KEY env)  --name NAME                 Server name (default: capsolver)

Programmatic

python
from capsolver_mcp.server import create_server
server = create_server(    api_key="your-key",       # or set CAPSOLVER_API_KEY env var    server_name="capsolver",  # name advertised to MCP clients    host="127.0.0.1",         # bind host for SSE / HTTP transports    port=8000,                # bind port for SSE / HTTP transports)server.run(transport="sse")   # or "stdio" or "streamable-http"

Note: host and port are constructor parameters on create_server() (forwarded to FastMCP), matching the MCP Python SDK 1.x API.

Configure in Claude Desktop

Add to your claude_desktop_config.json:

json
{  "mcpServers": {    "capsolver": {      "command": "capsolver-mcp",      "env": {        "CAPSOLVER_API_KEY": "your-key"      }    }  }}

To run without installing it globally, use "command": "uvx" with "args": ["capsolver-mcp"] — this is what MCP clients generate from the registry entry. See docs/mcp-integration.md for per-client examples.

Available tools

ToolBrowser?Description
solve_captchaNoSolve a captcha by type + site params (token mode)
detect_captchasYesScan a page URL and list present captcha types
solve_on_pageYesDetect + solve + autofill all captchas on a page
get_balanceNoCheck account balance and packages
get_supported_captchasNoList all supported captcha types and handlers

Browser-based tools (detect_captchas, solve_on_page) require the browser extra:

bash
pip install capsolver-mcp[browser]playwright install chromium

Development

bash
git clone https://github.com/capsolver-ai/capsolver-mcp.gitcd capsolver-mcpuv sync --all-extras          # or: pip install -r requirements-dev.txtuv run pytest                 # run testsuv run ruff check src tests   # lint

License

MIT

Source: README.md at commit 40a16f1

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

2
  1. v0.1.3LatestSep 18, 2026
  2. v0.1.2Sep 17, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts