Case 01076
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_01076Assembled 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
Fetch any URL and produce a structured content summary with extractive summarization, AI enhancement prompts, and structured output templates. Extracts clean text from HTML, identifies key sentences, generates metadata, and provides ready-to-use prompts for AI summarization. Perfect for content repurposing, research pipelines, and newsletter workflows.
SKILL.md
5.1 KB, ~1.2k tokens by cl100k_base, as published. Nobody here has run it
Content Summarizer
Transform any URL or raw text into structured, actionable content summaries using extractive summarization + AI enhancement prompts.
What This Skill Does
- Fetches any URL and extracts clean text (removes HTML, nav, scripts, ads)
- Extracts key sentences algorithmically (extractive summarization — no API needed)
- Generates metadata (title, word count, reading time)
- Outputs a structured markdown template + AI enhancement prompt ready for any LLM
The AI enhancement prompt makes the extracted content easy to summarize with MiniMax, GPT-4o, or Claude.
Quick Start
# Summarize a URL (outputs to terminal + /tmp/summaries/)
./scripts/url-to-summary.sh https://example.com/article
# Summarize text directly (from argument or pipe)
./scripts/summarize.sh "Your article text here..."
cat article.txt | ./scripts/summarize.sh
# Custom model for AI prompt
./scripts/url-to-summary.sh https://example.com --model openai
Scripts
url-to-summary.sh — Full Summary from URL
./scripts/url-to-summary.sh <url> [--model minimax|openai|anthropic]
Fetches URL, extracts clean text, produces:
- Extracted key sentences (extractive summarization)
- Structured markdown template (summary, tweet hook, key takeaways)
- AI enhancement prompt — copy-paste into any LLM for full summarization
- Metadata (title, word count, reading time)
Output saved to /tmp/summaries/summary-TIMESTAMP.md
summarize.sh — Summarize Text Directly
./scripts/summarize.sh "Your text..."
cat file.txt | ./scripts/summarize.sh
Same output as url-to-summary.sh but for raw text input.
Output Format
# Content Summary
**Source:** [url]
**Title:** [extracted or unknown]
**Stats:** ~1,200 words | ~6 min read
## Key Sentences (Extractive Summary)
[Top 15-20 most meaningful sentences from the text]
## Structured Summary Template
[TODO fields for: what, why, key points]
## Tweet Hook
[TODO: Compelling one-liner]
## Key Takeaways
1. [TODO]
2. [TODO]
3. [TODO]
## AI Enhancement Prompt
[Ready-to-use prompt with extracted text embedded]
Environment Variables
| Variable | Default | Description |
|---|---|---|
OUTPUT_DIR | /tmp/summaries | Where to save summary files |
MODEL | minimax | AI model for enhancement prompt |
Use Cases
- Content repurposing: Fetch articles, extract key points, feed into content pipeline
- Research pipeline: Batch summarize 20 URLs, collect key points, synthesize insights
- Newsletter workflow: Fetch sources, generate structured summaries, assemble newsletter
- Quick reading: Extract key sentences from long articles for rapid triage
Notes
- Uses extractive summarization (algorithmically selects important sentences) — no API key needed
- AI enhancement prompt is the value-add: paste into MiniMax/GPT/Claude for full abstractive summary
- HTML extraction removes scripts, styles, nav, footer, ads — focuses on article body
- Limited to 15,000 characters of extracted text to keep context manageable
Output Format
## Summary
[2-3 paragraph summary of the content]
## Tweet Hook
> [One compelling, shareable sentence that captures the essence]
## Key Takeaways
1. [First key point]
2. [Second key point]
3. [Third key point]
---
*Source: [URL] | ~X min read | ~Y words*
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
SUMMARIZER_MODEL | No | auto | LLM model for summarization (or let the agent auto-select) |
MAX_TOKENS | No | 500 | Max tokens in summary output |
Usage Examples
# Summarize a blog post
./scripts/url-to-summary.sh https://yourcompetitor.com/blog/industry-trends
# Process a research paper
./scripts/summarize.sh "$(curl -sL https://arxiv.org/pdf/2301.00001.pdf | head -c 10000)"
# Quick key points for a meeting
./scripts/key-points.sh https://news.ycombinator.com/item?id=12345678
# Use with OpenClaw: feed the output directly into your workflow
SUMMARY=$(./scripts/url-to-summary.sh https://example.com/article)
echo "$SUMMARY" | ./scripts/key-points.sh
Tips
- For better results, provide clean article URLs (avoid paywalls and heavy JS pages)
- Use
MAX_TOKENS=1000for longer, more detailed summaries - Chain with other skills: competitor-tracker alerts → content-summarizer → social posting
- The tweet hook is designed to be engaging, not just a truncated sentence
- Key points should be actionable insights, not just topic labels
Dependencies
curl— for fetching URLs- OpenClaw agent with web_fetch tool for content extraction (used internally by the agent)