Hugo ask

Ask ChatGPT or Claude about tickets that live on-chain.

Hugo ask stores 500 customer-support tickets in a canister on an OpenCloud cloud engine. That canister is the source of truth. This is a non-commercial demo for learning. You ask in ChatGPT or Claude. The ICP MCP connector works in both. It reads this site, then queries the canister.

A green card catalog with one drawer open, holding a row of ticket cards.
01
You add the ICP MCP connector in ChatGPT or Claude. After it is signed in, you give the agent this site’s frontend canister id.
02
The connector fetches /.well-known/ic-architecture from this site. That file names the frontend and the backend.
03
It reads the backend’s Candid interface and calls getApiDoc, then schema.
04
For your question it calls execute with a JSON query. The canister on the OpenCloud engine answers from its own rows: it filters, counts, or returns a page of tickets. It does not call an AI model.

This website does not answer questions. On Access, an admin signs in and turns principals on. ChatGPT or Claude queries the canister as the Internet Identity for this site, and only an enabled principal receives rows. The longer account is the project.

Hugo ask is a non-commercial demo for learning. The tickets stay on an Internet Computer canister. An admin enables who may query them. The ICP MCP connector, in ChatGPT or Claude, asks the questions.