Skip to content

Web search & research

Out of the box Pi can’t browse the web. These extensions add web search, URL fetching, and richer research so the model can ground its answers in current information.

A broad web toolkit: web search, URL fetching, GitHub repo cloning, PDF extraction, YouTube video understanding, and local video analysis.

Terminal window
pi install npm:pi-web-access

Benefits

  • The most popular all-in-one option — covers most research needs.
  • Goes beyond search: PDFs, repos, and video.

Drawbacks

  • Large surface area; you may only use a fraction of it.
  • Some features depend on external API keys/services.

Web search and fetch with pluggable providers — Brave, Tavily, Serper, Exa, Jina, Firecrawl, SearXNG, and Ollama — so you can choose your search backend.

Terminal window
pi install npm:@juicesharp/rpiv-web-tools

Benefits

  • Provider-agnostic: pick the search engine you already pay for (or self-host SearXNG).
  • Part of the broader rpiv-* family that shares conventions.

Drawbacks

  • You must configure a provider and its API key.
  • More moving parts than a single-provider tool.

Web search and fetch backed by Ollama’s web search and fetch APIs — a good fit if you’re already in the Ollama ecosystem.

Terminal window
pi install npm:@ollama/pi-web-search

Benefits

  • Simple, single-provider setup for Ollama users.
  • Maintained by the Ollama team.

Drawbacks

  • Tied to Ollama’s web APIs.
  • Fewer extras than the broader toolkits.
  • Want everything in one package: pi-web-access.
  • Want to choose/self-host your search backend: @juicesharp/rpiv-web-tools.
  • Already use Ollama: @ollama/pi-web-search.