pi-webaio

All-in-one web tools for pi with search (Google, Brave, DDG) and fetch with headless browser AI summarization

Packages

Package details

extension

Install pi-webaio from npm and Pi will load the resources declared by the package manifest.

npm repohome report
$ pi install npm:pi-webaio
Package
pi-webaio
Version
0.7.1
Published
Jul 20, 2026
Downloads
1,271/mo · 271/wk
Author
apmantza
License
MIT
Types
extension
Size
999.6 KB
Dependencies
13 dependencies · 1 peer
Pi manifest JSON
{
  "extensions": [
    "./pi-entry.mjs"
  ]
}

Security note

Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.

README

pi-webaio

pi-webaio

All-in-one web access tools for pi: search, fetch, crawl, extract, map, cache, chunk, and render web content for AI agents.

What It Does

pi-webaio registers eight pi tools:

  • aio-websearch — search DuckDuckGo, Brave, Yahoo, Bing, and Google in parallel
  • aio-webfetch — fetch one or many URLs into markdown or structured formats
  • aio-webcontent — retrieve cached content by URL
  • aio-webresult — retrieve cached results by response ID
  • aio-webmap — discover site pages or map GitHub repositories without fetching
  • aio-webpull — crawl/pull sites into local markdown files
  • aio-webquery — BM25 search over a locally-pulled corpus (offline, no re-fetching)
  • aio-webresearch — single-round research bundle: search → rank → fetch → cited evidence bundle on disk

It includes anti-bot TLS fingerprinting, browser fallback, GitHub/YouTube/package registry extractors, RAG chunking, TUI progress rendering, phase-aware errors, and opt-in paywall bypass support.

Install

pi install npm:pi-webaio

Or from git:

pi install git:github.com/apmantza/pi-webaio

Documentation

  • Features — overview, extraction pipeline, GitHub/YouTube handling, output formats, chunking, errors, and search ranking
  • Custom vertical extractors — add your own site extractors (company wikis, niche sites) without forking
  • Usage guide — common pi prompts and examples
  • Tools reference — tool names, parameters, and defaults
  • Architecture — build, TUI rendering, FetchError system, CI, and security notes
  • MCP server — use the tools from Claude Code, Claude Desktop, and other MCP clients without pi (npx -y pi-webaio-mcp)
  • PageMap inspiration — future extraction and structured-output ideas

Contributing

We especially welcome contributors for new vertical extractors, search engines, site-specific fetch fixes, anti-bot/paywall resilience, and docs. See CONTRIBUTING.md for setup, testing, and contribution checklists.

Contributors

Thanks goes to these wonderful people:

If you land a pull request or report an issue that gets fixed, we'll add you here.

License

pi-webaio is released under the MIT License.