Skip to main content
Any client that can reach a remote MCP server over streamable HTTP can use PostQueen. Whatever the client calls its fields, it needs the same three things: a name for the server, the address, and your API key, either in the address or in a header. The agents list has a guide for each client PostQueen covers.
Where the API key is: in the app, open Connections > API key. Only a workspace Admin or Super Admin can reveal or rotate it; other members see it masked. A workspace has one key, and each agent card on Connections shows your MCP address with the key already filled in.

​
The addresses

Clients that still use the older SSE transport can connect to https://api.postqueen.ai/sse/YOUR_API_KEY.
docs.postqueen.ai/mcp is a different server: it searches these docs and cannot post. Use an address from the table above.
Start with the key address. It needs no header and no sign-in, so it works in almost every client. Use the Bearer header when the client has a header field and you want the key out of the address. The sign-in address works only in clients that support MCP sign-in.

​
Config shapes

Most clients take the server in one of these shapes. Check your client’s docs for which one it reads.
The first shape is the one Cursor and Devin Desktop use, and the second is a VS Code project’s .vscode/mcp.json (VS Code’s MCP: Add Server now writes mcpServers with "type": "http"). Cline needs "type": "streamableHttp": without it, Cline falls back to the older SSE transport. Zed reads context_servers, Gemini CLI writes url with "type": "http", and Muse Code needs a schema_version. Their pages have the exact blocks: Cursor, VS Code, Devin Desktop, Zed, Gemini CLI, Muse Code. If the client asks for a transport, choose streamable HTTP. Choose SSE only when the client offers nothing else, and then use https://api.postqueen.ai/sse/YOUR_API_KEY.
Claude Desktop does not read a remote address from claude_desktop_config.json. Add PostQueen in Claude as a custom connector instead: see Claude.

​
Clients that only run local servers

Some clients can only start a program on your computer and talk to it over standard input and output (stdio). Bridge them to PostQueen with mcp-remote, which runs on your computer and forwards the calls. It needs Node.js.
Cursor, Codex CLI and Claude Desktop on Windows break a value with a space inside args. So in the Bearer header form, the header value comes from env, with no space after Authorization:. Use the bridge only when the client cannot take an address. A direct connection has fewer parts that can fail.

​
Keep the key safe

The key can post to every channel in your workspace. It sits in the client’s config file either way, so keep that file out of version control and off shared screens. If the key leaks, an admin chooses Rotate key under Connections > API key, and you update every client that used it.

​
Check that it works

Ask the client’s agent:
It calls integrationList and answers with your connected channels. If the connection fails or no PostQueen tools appear, see the agent will not connect over MCP.

​
Next steps

Every agent

Which agents can post to PostQueen, and how each one connects.

The MCP tools

What each of the 25 tools does, and what an agent cannot do.

Prompts and tool calls

The order an agent calls the tools in, with prompts that work.

Troubleshooting agents

What to do when a connection or a command fails.
Last modified on October 21, 2026