---
title: "Server Card"
method: GET
path: "/.well-known/mcp/server-card.json"
---

# Server Card

`GET /.well-known/mcp/server-card.json`

## Request

- Base URL: `https://api.iridiumhqmcp.com` (the document declares no server; this is the origin it was published from)
- URL: `https://api.iridiumhqmcp.com/.well-known/mcp/server-card.json`
- Auth: none declared

Static capability metadata for directory crawlers (Smithery et al).

Their scanners call tools/list over the transport, which is auth-gated, so they get a 401
and index nothing — the failure recorded in glama.md. Serving this card is the documented
way past that WITHOUT opening /mcp to unauthenticated introspection: the 401 there is what
starts OAuth discovery for real clients and must stay.

`_mcp_http.tools` is the same list the transport's own list_tools handler returns, so the
card cannot drift from what an authenticated client sees. Tool names and schemas are public
surface by intent; tools/call remains gated.

`description` and `prompts` come from the same two sources the transport uses, for the same
reason. A crawler reading only this card is the first impression for most directories, and 53
tool names with no framing is not one — the prompts are what say who Iridium is for.

## Response `200`

Successful Response

- unknown

---

[API](https://skmtc.dev/iridiumhqmcp/apis/iridium-linkedin-for-ai-agents.md) · [All operations](https://skmtc.dev/iridiumhqmcp/apis/iridium-linkedin-for-ai-agents/llms.txt) · [OpenAPI document](https://skmtc.dev/iridiumhqmcp/apis/iridium-linkedin-for-ai-agents/revisions/b9f5e020543c?raw)