Case 04325
A fast, offline static risk analysis CLI for AI agent skill files. Detects malicious instructions, steganographic payloads, and dangerous capability chains.
npx -y skills add knownasnaffy/prompthound --skill case_04325Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Search using a custom SearXNG instance via HTTP API. Enables privacy-friendly web search by aggregating results from multiple search engines. Supports query, format (json/csv/rss), language, time range, categories, engines, pagination, and safe search filtering. Requires a user-provided or configured SearXNG instance URL.
SKILL.md
5.8 KB, as published. Nobody here has run it
SearXNG Search
This skill enables web search using a custom SearXNG instance via its HTTP API.
Overview
SearXNG is a privacy-respecting metasearch engine that aggregates results from multiple search engines. This skill allows you to search using a custom SearXNG instance URL.
API Endpoints
GET /orGET /search- Query parameters in URLPOST /orPOST /search- Form data
Required Configuration
You need a SearXNG instance URL. The user can provide:
- A public instance URL (e.g.,
https://searx.be) - A self-hosted instance URL
- An environment variable
SEARXNG_URLcontaining the instance URL
Parameters
| Parameter | Required | Description |
|---|---|---|
q | Yes | Search query string |
format | No | Output format: json, csv, rss (default: HTML) |
language | No | Language code (e.g., en, zh, de) |
pageno | No | Page number (default: 1) |
time_range | No | Time filter: day, month, year |
categories | No | Comma-separated category list |
engines | No | Comma-separated engine list |
safesearch | No | Safe search level: 0, 1, 2 |
Categories and Engines
SearXNG organizes search into categories (tabs) with multiple engines per category.
Available Categories:
general- General web searchimages- Image searchvideos- Video searchnews- News articlesmap- Maps and locationsmusic- Music and audioit- IT/Technology resourcesscience- Scientific publicationsfiles- Files and torrentssocial_media- Social media content
Common Engines:
- Web:
google,duckduckgo,bing,brave,startpage,wikipedia - Images:
google images,bing images,unsplash,flickr - Videos:
youtube,google videos,vimeo,dailymotion - News:
google news,bing news,reuters - IT:
github,stackoverflow,arch linux wiki,pypi - Science:
google scholar,arxiv,pubmed
Bang Syntax: Use ! prefix in queries to target specific engines:
!go python- Search Google!wp artificial intelligence- Search Wikipedia!github machine learning- Search GitHub
For detailed engine lists and features, see references/engines_and_categories.md.
Usage
Using the Python Script
# Basic search
python scripts/searxng_search.py -u https://searx.example.org -q "python tutorial"
# JSON output
python scripts/searxng_search.py -u https://searx.example.org -q "python tutorial" --format json
# With language and time range
python scripts/searxng_search.py -u https://searx.example.org -q "news" --lang en --time-range day
# Use environment variable for URL
export SEARXNG_URL=https://searx.example.org
python scripts/searxng_search.py -q "search query"
# Search specific category
python scripts/searxng_search.py -u https://searx.example.org -q "AI" --categories news
# Search specific engines
python scripts/searxng_search.py -u https://searx.example.org -q "python" --engines google,stackoverflow
Using HTTP Requests Directly
import requests
url = "https://searx.example.org/search"
params = {
"q": "python tutorial",
"format": "json",
"language": "en",
"categories": "general,it", # Multiple categories
"engines": "google,duckduckgo" # Specific engines
}
response = requests.get(url, params=params, timeout=30)
results = response.json()
Bang Syntax in Query
# Use bang syntax to target specific engines
params = {
"q": "!github machine learning framework", # Only search GitHub
"format": "json"
}
JSON Response Format
{
"query": "search query",
"number_of_results": 1000000,
"results": [
{
"title": "Result Title",
"url": "https://example.com",
"content": "Snippet text...",
"engine": "google",
"score": 1.0
}
],
"answers": [],
"suggestions": [],
"unresponsive_engines": []
}
Important Notes
- Format Availability: JSON/CSV/RSS formats must be enabled in the instance's
settings.yml. Many public instances disable these formats. - Rate Limiting: Be respectful of the instance's resources. Add delays between requests.
- Self-Hosting: For reliable API access, consider self-hosting a SearXNG instance.
- Time Range: Not all engines support time range filtering. Check instance preferences.
- Engine Availability: Not all engines are enabled on all instances. Check the instance's preferences page.
Example Workflows
Basic Web Search
- Ask user for SearXNG instance URL or use configured URL
- Construct search query
- Call API with
format=json - Parse and present results
Category-Specific Search
- Choose category based on query type (news, images, science, etc.)
- Use
categoriesparameter or bang syntax - Filter results as needed
Multi-Page Search
- Perform initial search
- If more results needed, increment
pagenoparameter - Combine results from multiple pages
Filtered Search
- Use
time_range=dayfor recent results - Use
safesearch=2for strict filtering - Use
languageparameter for locale-specific results - Use
categoriesorenginesto narrow search scope
References
- engines_and_categories.md - Complete list of engines and categories