The usual workflow with an AI assistant is fragile: export products, paste part of the file into the chat and hope the context is enough. That breaks as soon as the catalog is large or the question depends on current data.
With Girofeeds MCP, the agent connects to your account and queries the catalog through a controlled server. The result is not a one-off answer from a stale export; it is work done against the same data you manage in the panel.
What you can ask
List products blocked by feed rules or missing required attributes.
Ask which provider or category is generating more cleanup work.
Review titles, descriptions and images without exporting CSV files.
Let an agent prepare optimization proposals from live catalog data.
How the connection works
The server is available over HTTP. Clients that already understand HTTP MCP can connect directly; clients that only speak stdio can use mcp-remote as the bridge. Authentication is done with an API key or OAuth 2.1, depending on how you want to approve access.
Setup checklist
- Open Connections MCP in the Girofeeds panel and copy the server URL.
- Create an API key with only the scopes the agent needs.
- Add the server to your MCP-compatible client, using HTTP directly or mcp-remote for stdio clients.
- Ask a narrow first question and check that the answer matches what you see in the panel.
Why scopes matter
An agent should not receive more access than the task needs. Girofeeds scopes let you separate catalog, providers and optimization areas, rotate keys and revoke access from the panel when a workflow is finished.
From file handoffs to live catalog work
The practical change is simple: the agent can answer questions about blocked products, weak attributes or provider status without you preparing a file first.