Web search

Web search and page reading#

Agents search the web with web_search and read pages with web_fetch. Search needs at least one search engine with your own key; nothing is preconfigured.

Engines answer searches: Tavily, Exa, Brave and Serper.
Readers turn a web page into clean text: Jina and Firecrawl. Without a reader, pages are read directly.

Engines are tried in the order you arrange them. The first enabled engine that returns results answers; an engine that fails or finds nothing hands over to the next one. Each engine has its own key, an optional base URL and an on/off switch.

Reading a page on your own machine or network (localhost, private and link-local addresses) is allowed, but asks for approval unless the project is in Full access.

Setting it up in the terminal#

Open Settings (F2) at Search & web. The engines are listed in fallback order, each with its key and its switch; the readers are on the same page.

From a shell:

printf '%s' "$TAVILY_API_KEY" | ncode config secret search_provider:tavily --stdin
ncode config search enable tavily
ncode config search enable exa
ncode config search order tavily,exa
ncode config search disable brave

The engine names are tavily, exa, brave and serper. Keys are read from stdin only, and a key is never shown again: its row reads ●●●●●●●● set · ends ….

Fetching local addresses#

An agent may read pages on localhost or your private network (a dev server, an internal wiki), but in read-only and auto mode each such fetch asks you first, on an approval card. Public pages never ask.