agentsclimarketplace

Case 05553

Skill knownasnaffy/prompthound/dataset/case_05553

A fast, offline static risk analysis CLI for AI agent skill files. Detects malicious instructions, steganographic payloads, and dangerous capability chains.From the repository description

Install
npx -y skills add knownasnaffy/prompthound --skill case_05553

Assembled 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.

SKILL.md

2.1 KB, 463 tokens by cl100k_base, as published. Nobody here has run it

Snowflake Documentation Scraper

Scrapes docs.snowflake.com sections to Markdown with SQLite caching (7-day expiration).

Usage

First time setup (auto-installs uv and doc-scraper):

python3 .claude/skills/doc-scraper/scripts/doc_scraper.py

Subsequent runs:

doc-scraper --output-dir=./snowflake-docs
doc-scraper --output-dir=./snowflake-docs --base-path="/en/sql-reference/"
doc-scraper --output-dir=./snowflake-docs --spider-depth=2

Command Options

OptionDefaultDescription
--output-dirRequiredOutput directory for scraped docs
--base-path/en/migrations/URL section to scrape
--spider-depth1Link depth: 0=seeds, 1=+links, 2=+2nd
--limitNoneCap URLs (for testing)
--dry-run-Preview without writing

Output

output-dir/
├── SKILL.md              # Auto-generated index
├── scraper_config.yaml   # Editable config (auto-created)
├── .cache/               # SQLite cache (auto-managed)
└── en/migrations/*.md    # Scraped pages with frontmatter

Configuration

Auto-created at {output-dir}/scraper_config.yaml:

rate_limiting:
  max_concurrent_threads: 4
spider:
  max_pages: 1000
  allowed_paths: ["/en/"]
scraped_pages:
  expiration_days: 7

Troubleshooting

IssueSolution
Too many pagesLower --spider-depth or edit config
Missing pagesIncrease --spider-depth
Cache corruptionDelete {output-dir}/.cache/ (rare)

What ships with it: 6 files

50.2 KB alongside SKILL.md, 2 of them executable

scripts/

Gives 0 of the 12 instructions most scraping extraction skills give in 463 tokens

Counted across 109 of the 148 authors here whose files we hold, read 2026-09-06

  • Add delays between requestsin 7 of 109, across 6 files
  • Close the browser when finishedin 6 of 109, across 4 files
  • Authenticate every request with the x-api-key headerin 5 of 109, across 2 files
  • Default to Scraper APIs when one existsin 5 of 109, across 2 files
  • Map site structure before a full crawlin 5 of 109, across 3 files
  • Use the Web Scraping API for arbitrary URLsin 4 of 109, across 1 file
  • Treat data as valid only if requestMetadata.status is okin 4 of 109, across 1 file
  • Set client timeout to at least 300 secondsin 4 of 109, across 1 file
  • Retry only 429 and 5xx with exponential backoffin 4 of 109, across 1 file
  • Cap concurrency at your plan limitin 4 of 109, across 1 file
  • Persist the async job id immediatelyin 4 of 109, across 1 file
  • Poll job status every 10 to 30 secondsin 4 of 109, across 1 file

Said here and by no other author read

  • Run the setup script on first use
  • Pass --output-dir on every run
  • Set --base-path to scrape a specific section
  • Set --spider-depth to control link depth
  • Use --limit to cap URLs while testing
  • Use --dry-run to preview without writing

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.