Fetch URL

fetch_url reads one page and returns its text. Without it the agent can only answer from what the model already knows, which was fixed when the model was trained. With it, it can check what a specific page says today.

Usage

Have a look at https://go.dev/doc/go1.26 and tell me what changed in the garbage collector.

It reads and nothing more. It cannot log in, fill in a form or run a page's JavaScript, so a site that renders entirely in the browser comes back empty. Pages are cut to roughly 6,000 characters, because a whole page would crowd out the conversation itself.

It also refuses to connect to your own machine or your local network — localhost, 127.0.0.1, 10.x, 192.168.x and the rest, including a public hostname that resolves to one of them. Ollama, Redis and anything else you run stay out of reach, whoever is doing the asking.

web_search is the tool to reach for first when you don't already have a link — the agent calls fetch_url itself on whichever result looks worth opening.

The model needs tool support

Tool calling depends on the model. If LLM_MODEL has no tools capability, Mycel warns at startup and fetch_url is ignored along with the rest.