Connect to AI tools
After deploying your MCP server, connect it to your AI assistant so the assistant can call your selected MCP tools.
Build absolutely anything...with the right context.
| Audience | MCP server owners and developers connecting AI clients |
|---|---|
| Prerequisites | A deployed MCP server, copied config snippet, and access to Cursor, Claude Desktop, Windsurf, or another MCP client |
The config snippet
After deployment, you receive a JSON snippet:
{
"mcpServers": {
"my-docs": {
"url": "https://appatools.com/mcp-studio/api/mcp/my-docs"
}
}
}
Use streamable HTTP (/api/mcp/{slug}) as the default endpoint. Keep SSE (/api/mcp/{slug}/sse) only when a specific MCP client still requires it.
Supported clients
Cursor
- Open Settings → MCP Servers
- Paste the JSON config
- Restart Cursor
Claude Desktop
- Open Settings → Developer → Edit Config
- Add your server to the
mcpServersobject - Restart Claude Desktop
Windsurf
- Open your MCP configuration file
- Add the server entry
- Restart Windsurf
Other MCP clients
Any client that supports the MCP standard can connect using the streamable HTTP endpoint URL. This includes VS Code extensions, JetBrains plugins, Neovim, and custom integrations. If a legacy client still expects SSE, use /api/mcp/{slug}/sse.
Verify the connection
After connection, your AI tool should list the MCP tools you selected, such as search_docs and get_code_examples. Ask a question related to your indexed sources. The AI agent should call your MCP tools during the conversation.
What if it's not working?
- Make sure you copied the full JSON snippet, including the outer
mcpServerswrapper. - Check that your MCP server's status is Active on the Dashboard.
- Restart your AI tool after adding the config.
- If sources are still indexing, the server works but might use live fetching, which can be slower. Indexing continues in the background even if you close the dashboard.