LangChain Integration

from langchain_core.tools import tool
from gatesolve import GateSolve

gs = GateSolve(api_key="gs_your_key")

@tool
def solve_captcha(captcha_type: str, site_key: str, page_url: str) -> str:
    """Solve a CAPTCHA and return the solution token.
    
    Args:
        captcha_type: Type of CAPTCHA (turnstile, recaptcha_v2, recaptcha_v3, hcaptcha)
        site_key: The site key from the CAPTCHA element
        page_url: URL of the page containing the CAPTCHA
    """
    solution = gs.solve(
        captcha_type=captcha_type,
        sitekey=site_key,
        pageurl=page_url,
    )
    return solution.token

from langchain_openai import ChatOpenAI
from langchain.agents import AgentExecutor, create_tool_calling_agent
from langchain_core.prompts import ChatPromptTemplate

# Create the agent with the CAPTCHA solving tool
llm = ChatOpenAI(model="gpt-4o")

prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a web automation agent. When you encounter a CAPTCHA, "
     "use the solve_captcha tool to get a token, then inject it into the page."),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])

agent = create_tool_calling_agent(llm, [solve_captcha], prompt)
executor = AgentExecutor(agent=agent, tools=[solve_captcha], verbose=True)

# The agent can now solve CAPTCHAs autonomously
result = executor.invoke({
    "input": "Go to https://example.com/login, solve any CAPTCHAs, and log in."
})

from langchain_core.tools import tool
from gatesolve import AsyncGateSolve

gs = AsyncGateSolve(api_key="gs_your_key")

@tool
async def solve_captcha_async(captcha_type: str, site_key: str, page_url: str) -> str:
    """Solve a CAPTCHA asynchronously and return the solution token."""
    solution = await gs.solve(
        captcha_type=captcha_type,
        sitekey=site_key,
        pageurl=page_url,
    )
    return solution.token

from langchain_core.tools import tool
from gatesolve import GateSolve

gs = GateSolve(api_key="gs_your_key")

@tool
def submit_captcha(captcha_type: str, site_key: str, page_url: str) -> str:
    """Submit a CAPTCHA for solving without blocking. Returns a job ID.
    Use poll_captcha to check the result."""
    job_id = gs.submit(
        captcha_type=captcha_type,
        sitekey=site_key,
        pageurl=page_url,
        callback_url="https://your-server.com/webhook/captcha",
    )
    return f"Submitted. Job ID: {job_id}"

@tool
def poll_captcha(job_id: str) -> str:
    """Check the status of a CAPTCHA solve job."""
    result = gs.poll(job_id)
    if result["status"] == "solved":
        return f"Solved! Token: {result['token']}"
    return f"Status: {result['status']}. Try again in a few seconds."

# Add GateSolve MCP server to your LangChain MCP config
# No Python code needed — just configure the MCP server

# In your MCP config (e.g., mcp_config.json):
{
  "servers": {
    "gatesolve": {
      "command": "npx",
      "args": ["@gatesolve/mcp-server"],
      "env": {
        "GATESOLVE_API_KEY": "gs_your_key"
      }
    }
  }
}

# LangChain agents using MCP will automatically discover
# the solve_captcha tool from the GateSolve MCP server