Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.enrichley.io/llms.txt

Use this file to discover all available pages before exploring further.

If something isn’t working, start with the quick diagnostic below. Most issues fall into one of four categories: connection, authentication, client-specific, or tool errors.

​
Quick diagnostic

Ask your AI assistant:
“Check my Enrichley credit balance.”
This calls enrichley_get_account_status, which is free and read-only.
  • Returns your credits: the connection and authentication are healthy. If a different tool isn’t working, it’s likely a tool-specific issue.
  • Returns an authentication error: see Authentication errors below.
  • Hangs or returns a connection error: see Connection issues below.

​
Connection issues

Confirm the URL is exactly https://mcp.enrichley.io/mcp. Trailing slashes, missing /mcp, or substituting the base URL will all fail.Test the health endpoint in a browser or with curl:
curl https://mcp.enrichley.io/health
A healthy server returns { "ok": true, "service": "enrichley-mcp-worker" }.
Some clients suppress popups. Look for a clickable link in the client’s terminal or status bar that you can open manually. If you have a popup blocker, disable it for the client and retry.
The redirect URL your client is using may not match what the server expects. Reset the client’s MCP configuration, remove and re-add the Enrichley server, and try again.

​
Authentication errors

The signed-in account does not have an active Enrichley plan or a ready API credential. Sign in with the right account or upgrade your plan, then reconnect the client.
The account API credential is missing, disabled, or temporarily unavailable. Open your Enrichley profile to confirm the account is ready, then reconnect the client.
Your AI client’s OAuth token has expired. Reconnect to refresh it.
  • Claude Code: run /mcp, clear authentication for enrichley, then reconnect.
  • Cursor: disable and re-enable the Enrichley MCP server.
The client is sending requests without a valid token. Run the setup flow again from scratch for that client.

​
Client-specific issues

Run claude mcp list to confirm enrichley is registered. If it isn’t, re-run the install command from Setup. After install, run /mcp to start the OAuth flow — the server only becomes usable once you complete sign-in.
Restart Cursor after editing mcp.json. Open the MCP/tools panel and confirm the Enrichley server is enabled. If it’s enabled but tools are missing, complete the OAuth browser flow when Cursor prompts you.
Make sure your config uses "servers" (not "mcpServers") and includes "type": "http". After saving, run MCP: List Servers from the command palette and select enrichley to start the OAuth flow.
The server is registered but you haven’t completed OAuth yet. Run codex mcp login enrichley and complete the browser sign-in.

​
Tool errors

The enrichley_validate_email tool requires a syntactically valid email. If your AI assistant pulled an address from text, double-check it for typos or missing characters.
You’ve hit the 10 requests/second limit. Wait until rateLimitResetAt (returned in the previous tool’s usage envelope) and retry. For higher throughput, contact support@enrichley.io.
The validation tool consumes one credit per call. Top up or upgrade your plan to keep validating.
result: "unknown" means the upstream provider could not determine the address’s status — usually because the receiving server didn’t respond. Treat the address as not-safe-to-send and retry later if needed.

​
FAQ

No. Streamable HTTP only.
No. The key is stored encrypted on the server. Your AI client only ever holds a short-lived OAuth token.
The fastest way is to disconnect inside the client itself (/mcp in Claude Code, the MCP panel in Cursor). To revoke access for every client at once, rotate your API key on your Enrichley profile.
The Enrichley MCP server implements OAuth 2.1 with PKCE, which is what ChatGPT’s developer-mode MCP support requires. We haven’t validated the integration end-to-end yet, so we’re not publishing official setup steps until we do. If you try it and it works, let us know at support@enrichley.io.
No self-hosted version is available. The hosted server at https://mcp.enrichley.io/mcp is the only supported deployment.

​
Debugging with MCP Inspector

MCP Inspector is the official open-source debugging tool from the Model Context Protocol team. Use it to test the Enrichley server outside of an AI client — it’s especially helpful when you want to confirm whether an issue is on the server or on your client’s MCP integration.

​
Start the inspector

npx @modelcontextprotocol/inspector
This starts a local UI in your browser.

​
Connect to Enrichley

In the inspector UI, configure:
  • Transport: Streamable HTTP
  • Connection type: Via Proxy
  • URL: https://mcp.enrichley.io/mcp
  • Scope: account.read validation.run
  • Custom Authorization header: disabled
Click Connect, complete the browser sign-in, and authorize the requested scopes. The inspector will list the two tools and let you call them interactively.

​
Common inspector gotchas

Confirm the transport is set to Streamable HTTP (not SSE) and the URL ends with /mcp. Some inspector versions default to other transports or other paths.
The inspector relies on your default browser handling the OAuth redirect. If you have multiple browsers installed, set your preferred one as the system default before retrying.
Refresh the inspector’s tool list by reconnecting. If they still don’t show, check the inspector’s network tab for a 401 response — that points to an authorization issue, not a tool issue.

​
Still need help?

Email support

Contact our team for account or billing issues.

API Reference

Full REST API documentation with an interactive playground.