Connect ChatGPT
How to use Qyvo with ChatGPT's MCP integration when it ships.
OpenAI announced MCP client support in late 2025; rollout to ChatGPT (web + desktop) and the Responses API is incremental. This page describes the connection shape Qyvo expects — when MCP is enabled in your ChatGPT plan, the steps below should apply directly.
Generic config
Qyvo ships a standard MCP HTTP transport. ChatGPT's MCP UI asks for the same fields as every other client:
| Field | Value |
|---|---|
| Server URL | https://www.qyvo.io/mcp |
| Transport | HTTP |
| Auth header | Authorization: Bearer YOUR_TOKEN_HERE |
| Server name (display) | Qyvo |
Generate the token in Settings → API Tokens, then paste it into ChatGPT's MCP server form.
Until MCP ships in ChatGPT
If you want LLM control of your Qyvo workspace today and you're an OpenAI user, the supported paths are:
- OpenAI Responses API + MCP — the API has shipped MCP support; you can wire Qyvo as a tool source from your own application code.
- Claude or Cursor — both have shipped MCP for months. See the dedicated pages.
- n8n's AI Agent — works with any LLM provider; install
n8n-nodes-qyvoand chain the action node under any agent.
Permissions reminder
ChatGPT's planned UI will prompt for tool approval on first use of each tool, like Claude and Cursor. Read-only tools (list_*, get_*) will auto-approve once you grant them. Write tools will prompt every time unless you opt into "always allow" per tool.