| name | websearch |
| description | Searches the web via DuckDuckGo and returns results as markdown, CSV, or JSON. Use this skill whenever the user wants to search the web, look up information online, find URLs, do web search, research a topic, or needs current/web-based answers. |
| metadata | {"tags":["meta"]} |
websearch
Search the web via DuckDuckGo and return structured results.
Overview
websearch fetches search results from DuckDuckGo, parses the HTML response, and outputs them in a structured format.
Supports three output formats: markdown (default), CSV, and JSON. Results can be written to stdout or a file.
Usage
websearch.sh "react hooks tutorial"
websearch.sh "python async patterns" --format json --output results.json
websearch.sh "rust vs go 2025" --format csv
websearch.sh --help
Options
| Flag | Default | Description |
|---|
--format | markdown | Output format: markdown, csv, json |
--output | stdout | Output destination: stdout or a file path |
Gotchas
- DuckDuckGo blocks automated requests — the script impersonates Safari (iPhone/Mac user agents) to reduce blocking. If results are empty, the site may have rate-limited the IP.
- URLs in results are DuckDuckGo redirect links — the
uddg parameter contains the actual URL (URL-encoded). The parser extracts and decodes it automatically.
- Results are limited to one page (10 results) — pagination is not supported. Use more specific queries to surface relevant results on the first page.
