capsolver-agent

Agent integrations for CapSolver — framework-agnostic tool definitions and LangChain BaseTool implementations.

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

For framework integration guides (OpenAI, LangChain, LlamaIndex, CrewAI, Google ADK, and more), see docs/agent-integration.md.

Install

pip install capsolver-agent
pip install capsolver-agent[langchain]   # with LangChain support
pip install capsolver-agent[browser]     # with Playwright support (for detect/solve_on_page)

All packages read the API key from the environment:

# bash / zsh
export CAPSOLVER_API_KEY="your-capsolver-api-key"

# PowerShell
$env:CAPSOLVER_API_KEY = "your-capsolver-api-key"

# cmd
set CAPSOLVER_API_KEY=your-capsolver-api-key

Framework-agnostic tools (any LLM / agent framework)

Use schema.py to get tool schemas as JSON and an async executor to run tool calls. Works with OpenAI function calling, OpenAI Agents SDK, Browser Use, or any custom agent loop.

import asyncio
from capsolver_agent.schema import get_all_tools, create_executor

async def main():
    # 1. Get tool schemas — feed to your LLM's function-calling API
    tools = get_all_tools()
    openai_functions = [t.to_openai_function() for t in tools]

    # 2. Execute a tool call returned by the LLM
    executor = create_executor(api_key="YOUR_API_KEY")
    result = await executor.execute("solve_captcha", {
        "captcha_type": "reCaptchaV2",
        "website_url": "https://example.com",
        "website_key": "6Le-wvkSAAAAAPBMRT...",
    })
    print(result)
    # {"success": True, "solution": {"token": "03AF...", ...}}

asyncio.run(main())

Each ToolDef provides two export formats:

tool = get_all_tools()[0]
tool.to_openai_function()  # → OpenAI function-calling schema
tool.to_json_schema()      # → MCP-style tool descriptor (name + inputSchema)

For a quick one-shot call without creating an executor:

from capsolver_agent.schema import execute_tool

result = await execute_tool("solve_captcha", {
    "captcha_type": "reCaptchaV2",
    "website_url": "https://example.com",
    "website_key": "6Le-wvkSAAAAAPBMRT...",
}, api_key="YOUR_API_KEY")

LangChain integration

Pre-built BaseTool subclasses with Pydantic input schemas — plug directly into any LangChain agent.

import asyncio
from capsolver_agent.langchain_tools import get_langchain_tools
from langchain_openai import ChatOpenAI
from langgraph.prebuilt import create_react_agent

tools = get_langchain_tools(api_key="YOUR_API_KEY")

llm = ChatOpenAI(model="gpt-4o")
agent = create_react_agent(llm, tools)

async def main():
    result = await agent.ainvoke({"messages": [...]})

asyncio.run(main())

Or import individual tools:

from capsolver_agent.langchain_tools import SolveCaptchaTool, GetBalanceTool

solver = SolveCaptchaTool(api_key="YOUR_API_KEY")
balance = GetBalanceTool(api_key="YOUR_API_KEY")

CLI

The capsolver-agent command lets you inspect available tools and their schemas from the terminal.

# List all tools with descriptions
capsolver-agent list

# Show JSON Schema for a specific tool
capsolver-agent schema solve_captcha

# Export all tools in OpenAI function-calling format
capsolver-agent schema --format openai

# Export one tool in OpenAI format
capsolver-agent schema --format openai detect_captchas

Also works via python -m capsolver_agent list.

Available tools

Browser-based tools require pip install capsolver-agent[browser] and playwright install chromium.

Integration examples

See the capsolver-ai-hub examples for runnable demos:

• openai_function_calling.py — agentic loop with OpenAI function calling
• openai_agents.py — OpenAI Agents SDK with @function_tool
• langchain_agent.py — LangChain ReAct agent
• browser_use_agent.py — Browser Use with @tools.action()

Development

git clone https://github.com/capsolver-ai/capsolver-agent.git
cd capsolver-agent
uv sync --all-extras          # or: pip install -r requirements-dev.txt
uv run pytest                 # run tests
uv run ruff check src tests   # lint

License

MIT