What the client needs
SEO Researcher exposes a remote Streamable HTTP server at:
https://mcp.seoresearcher.ai/mcpThe client must discover and invoke tools, complete OAuth with PKCE and the server’s discovery endpoints, and preserve its credentials. It must also support multiple turns: start a job, wait, check status, then retrieve results. A field labeled “API URL” or support for local STDIO alone is not enough.
OpenClaw and less common clients
Do not assume an OpenClaw extension, community bridge, or similarly named MCP integration supports this authentication flow. We have not verified a specific OpenClaw version and installation path. Consult the installed client’s official documentation for remote HTTP and OAuth support before configuring it. Do not copy a JSON shape from Cursor into a different client unchanged.
If OAuth is unsupported, ask support whether scoped API-key access is available for your account. Do not reuse a browser cookie or another product’s API key. A local bridge adds another dependency and credential store; it is not required for the native clients in our directory.
For developers
The tool reference describes the MCP lifecycle and REST equivalents. A custom REST integration needs its own authentication and error handling; an API URL is not an MCP URL.
Check that it works
Ask your agent:
Use SEO Researcher to list my most recent research job (limit 1).
Do not start research. Report whether the call succeeded, including
when the account has no saved jobs.A successful seo_research_list response—even an empty list—confirms authenticated access. Merely saving the URL or seeing a login screen does not. Then follow your first research workflow. If authorization does not return, use connection troubleshooting.