Getting started

Install

pi install git:github.com/jawwadzafar/pi-webfind
# or, after the npm publish:
pi install npm:pi-webfind

Pi loads the TypeScript extension directly — no build step, no config.

What you get

Seven tools and one command, all free:

Tool	Best for
web_search	news, articles, broad queries — set deep: true for excerpts
fetch_page	read a URL — pass query for relevant passages
search_stackoverflow	error messages, debugging
search_wikipedia	definitions, concepts, history
search_npm	JS/TS packages with quality scores
search_github	repos, stars, languages
search_hn	tech community opinion, launches

Plus the /research <topic> command: parallel search across web, HN, GitHub and Wikipedia, fetches the top pages, then briefs you with "documents worth reading + recurring conclusions".

Two habits that get better answers

Search deep, not wide. For factual questions pass deep: true — the top results get read and each row carries a query-relevant excerpt. Often the answer is on screen without a single fetch.

{ "query": "why does rust have lower latency than go", "max_results": 5, "deep": true }

Fetch with intent. Long pages return the passages that match your query, not the first 8000 characters:

{ "url": "https://en.wikipedia.org/wiki/DuckDuckGo", "query": "how does duckduckgo make money" }

Caches

Search results (10 min) and fetched pages (1 h) are cached under ~/.pi/agent/cache/webfind/ and survive restarts. refresh: true / no_cache: true skip them.

Security

pi-webfind runs with full system access like any pi extension. It makes outbound HTTPS requests to search engines, public APIs and pages you fetch. It never reads your files, never executes fetched content, and blocks requests to localhost/private networks (SSRF guard). Engine traffic uses a politeness throttle; the only optional credential is GITHUB_TOKEN to lift GitHub's 10 req/min anonymous limit.