~ cat ./projects/web-search/README.md

web-search

LIVE

Agent-friendly web search toolkit with fallbacks, caching, and evidence-first safer defaults.

#python#agents#search

Overview

Replace one-off search and scrape scripts with a single, agent-friendly toolkit. Designed to be used by AI coding agents that need to fetch information from the web.

Features

  • Multiple providers – DuckDuckGo (keyless), Brave API
  • Automatic fallbacks – if one provider fails, try another
  • Caching – avoid redundant requests
  • Evidence-first – returns source URLs alongside content
  • Safe defaults – respects robots.txt, rate limits

Usage

# Search
web-search query "next.js app router tutorial"

# Extract content from URL
web-search extract "https://example.com/article"

Why?

Agents frequently need to search the web or fetch page content. Rather than writing bespoke scripts each time, this provides a reliable, well-tested toolkit with sensible defaults.