Connect your agent
For anyone whose agent (Studio, Claude, Cursor or ChatGPT) should read these docs while it works. The endpoint is public, read-only, and needs no account or key:
https://docs.alphaagent.prometheusrl.com/mcp
Any client that supports remote MCP servers (streamable HTTP, JSON-RPC over POST) can use it.
AlphaAgent Studio
- Open Data Connectors, choose Create Connector and pick the type MCP Server.
- Set MCP Server URL to
https://docs.alphaagent.prometheusrl.com/mcp. Leave the credential rows empty. - Click Test Connection, then Discover Tools & Enrich Guide: Studio discovers the three tools and drafts a connector guide.
- Review the guide, click Accept & Use This Guide, then Create Connector (v1).
- Open the agent, choose Configure, then Add Connector, pick the new connector and Save & Activate. The agent now calls
searchandfetchwhen a question is about AlphaAgent.
Claude Code
claude mcp add --transport http alphaagent-docs https://docs.alphaagent.prometheusrl.com/mcp
Or commit .mcp.json at the project root so the whole team gets it:
{
"mcpServers": {
"alphaagent-docs": {
"type": "http",
"url": "https://docs.alphaagent.prometheusrl.com/mcp"
}
}
}
Run /mcp to confirm the server is connected, then ask, for example, "Using alphaagent-docs, how do I create a Snowflake connector?".
Claude Desktop and claude.ai
Open Settings, then Connectors, choose Add custom connector, enter the name AlphaAgent Docs and the URL above, leave authentication empty and save. It syncs to Claude Code on the same account.
Cursor
Add the server to ~/.cursor/mcp.json (all projects) or .cursor/mcp.json (one project):
{
"mcpServers": {
"alphaagent-docs": {
"url": "https://docs.alphaagent.prometheusrl.com/mcp"
}
}
}
Cursor lists the three tools under Settings, then MCP.
ChatGPT
- Open Settings, then Connectors, and enable Developer mode under Advanced settings.
- Choose Create, enter the name
AlphaAgent Docsand the URL above, and choose No authentication. - Enable the connector under the tools menu and ask. Citations link back to this site.
What the server can answer
| Tool | What it does |
|---|---|
search | A question or a few keywords in; the best-matching pages out as {id, title, url}. |
fetch | A page id from search in; the full page as Markdown out. |
list_sections | The table of contents: the eight sections in reading order, every page in each, and the published docs version. |
The index is rebuilt at every publish. Nothing you send is stored; requests are rate-limited per client.
Notes
- Plain-text copies for tools that read files:
https://docs.alphaagent.prometheusrl.com/llms.txt(every page, one line each),https://docs.alphaagent.prometheusrl.com/llms-full.txt(the full text of every page), and any page path with.mdappended, for examplehttps://docs.alphaagent.prometheusrl.com/studio-guide/data-connectors.md. - The server is stateless JSON-RPC over HTTPS. Only
POSTis accepted;GETreturns a JSON error rather than a long-lived stream. Aninitializecall looks like this;tools/listandtools/callfollow the same shape:
curl -s https://docs.alphaagent.prometheusrl.com/mcp \
-H 'Content-Type: application/json' \
-H 'Accept: application/json, text/event-stream' \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"curl","version":"1.0"}}}'
Related
- MCP connector: the general steps for connecting any MCP server to a Studio agent.