Connect Spider to any AI.
Spider's MCP server is hosted, so there is nothing to install and nothing to run locally. One URL, one approval, and a scoped key is minted for you. About a minute, start to finish.
Pick your client.
Everything below adapts to it.
Add the endpoint to Claude.
Paste the endpoint URL as a custom connector.
Open settingsApprove access.
Claude shows an OAuth prompt the first time it reaches the server. Approve it and a Spider API key is minted automatically. The key is scoped to this connector, revoke it anytime.
Connected.
Ask your AI to research something. It picks from 27 tools, crawl, scrape, search, and a real browser, while Spider handles proxies, browser sessions, and rendering.
Paste the config.
Any MCP client · add to its MCP config
{
"mcpServers": {
"spider": {
"type": "url",
"url": "https://mcp.spider.cloud/mcp"
}
}
}Setup by client, as text
Endpoint: https://mcp.spider.cloud/mcp. Transport: Streamable HTTP. The hosted server registers 27 tools. ChatGPT lists the seven core tools. Leave the key out of the Claude Code and Codex commands. The first tool call opens a browser approval that mints a scoped key. A header on the command skips that approval.
Claude Code
claude mcp add --transport http spider https://mcp.spider.cloud/mcpThen run /mcp in Claude Code, pick spider, and approve in the browser.
Codex
codex mcp add spider --url https://mcp.spider.cloud/mcp
codex mcp login spiderWith an API key instead of OAuth, for CI and servers
claude mcp add --transport http spider https://mcp.spider.cloud/mcp --header "Authorization: Bearer $SPIDER_API_KEY"
codex mcp add spider --url https://mcp.spider.cloud/mcp --bearer-token-env-var SPIDER_API_KEYOpenAI Agents SDK (Python)
A server process sends a key, since nobody is there to approve.
import asyncio, os
from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp
async def main():
async with MCPServerStreamableHttp(
name="spider",
params={
"url": "https://mcp.spider.cloud/mcp",
"headers": {"Authorization": f"Bearer {os.environ['SPIDER_API_KEY']}"},
},
) as spider:
agent = Agent(name="researcher", mcp_servers=[spider])
result = await Runner.run(agent, "Scrape https://example.com as markdown")
print(result.final_output)
asyncio.run(main())Any other MCP client
Add https://mcp.spider.cloud/mcp as a Streamable HTTP server. Approve the OAuth prompt on first connect, or send Authorization: Bearer with a key from https://spider.cloud/api-keys.