Skip to main content
Enterspeed hosts three Model Context Protocol servers. Point an AI client at one and it discovers a set of tools it can call on your behalf. You do not need to host anything yourself.

Query MCP

Ask questions about your data in Enterspeed using the MCP server for the Query API.

Management MCP

Inspect and configure your tenant — schemas, environments, sources, domains, and clients.

Documentation MCP

Search and retrieve documentation content or report back issues or suggestions.

Which one do you need?

Query MCP and Management MCP are complementary — many people connect both, so an agent can read a schema, change it, and then query the result.

How authentication works

Both tenant-facing servers work the same way: they are passthrough proxies that hold no credentials of their own. Your client sends a key on every request, the server forwards it unchanged, and the underlying Enterspeed API makes every authorisation decision. That means the key you issue is the whole boundary. For Query MCP, a key scoped to one environment and one index prefix gives an agent exactly that much and no more. For Management MCP, the key’s access level decides whether an agent can change anything at all.
Custom connectors on claude.ai in the browser are not supported yet. Browser-based custom connectors cannot send a custom HTTP header, and both tenant-facing servers authenticate with one. Use Claude Code, Claude Desktop, VS Code, Cursor, or your own agent instead.We are working on OAuth support to remove this restriction. Until it ships, a header-capable client is required. The Documentation MCP server needs no authentication and is unaffected.

Transport

Both tenant-facing servers speak MCP Streamable HTTP on the root URL. Responses are delivered as a server-sent event stream on that same URL, so clients that label their transport SSE generally work when pointed at the root. There is no separate /sse endpoint — always configure the root URL, and use the http transport type where your client offers a choice.