Case 05081

Skill knownasnaffy/prompthound/dataset/case_05081

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

Install

npx -y skills add knownasnaffy/prompthound --skill case_05081

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.

What its author says it does

Copied from the file, not written here

Use this skill when the user wants a long-form book summarized in the user's preferred language at an explicitly verified 20 percent compression ratio, with the result rejected if it falls outside the allowed range. Warning: this skill may consume many tokens because it can run multiple times until it reaches the requested compression rate.

SKILL.md

2.6 KB, 538 tokens by cl100k_base, as published. Nobody here has run it

Test Suite

This skill includes a comprehensive test suite. Test fixtures use mock credentials that are NOT real secrets:

# All tokens in tests/ are fake placeholders:
MOCK_API_KEY = 'sk-test-1234567890abcdef1234567890abcdef'
MOCK_AWS_KEY = 'AKIAIOSFODNN7EXAMPLE'

See tests/conftest.py for the full mock configuration.

Book Summarizer

Use this skill for requests like:

"Summarize this book at a 20% compression ratio"
"Generate a substantial summary and verify the ratio"
"Produce the summary in batches and validate the final total"

Rules

Default target ratio: 0.20
Default tolerance: 0.02
Accept only summaries between 18% and 22% of the original word count
Output must be in user's language of choice (e.g., pt-BR)
If the ratio is outside the allowed range, do not import the summary
In chat-only mode, if the source is very large, generate the summary in multiple batches and merge them before validation
All packaged helper scripts live in scripts/ inside this skill folder

Workflow

Start from a local plain-text book file or a downloaded Project Gutenberg text.
Count the source words with python scripts/book_tools.py count <original_file>.
Compute the target summary length with python scripts/book_tools.py target <original_file> --ratio 0.20.
If the source is too large for one reply, split it with python scripts/split_book.py <original_file> 3000.
Draft summary batches in order, preserving chronology and section fidelity.
Merge the batches with python scripts/book_tools.py aggregate <summary_file> <batch_files...>.
Validate the final ratio with python scripts/verify_summary_ratio.py <original_file> <summary_file>.

Key Files

scripts/book_tools.py
scripts/split_book.py
scripts/verify_summary_ratio.py
SKILL.md

Notes

The packaged scripts use only the Python standard library.
Run the commands from the skill folder, or use explicit paths if you call them from elsewhere.
For very large books, prefer the automated pipeline over single-turn chat drafting.
In chat-only mode, books above roughly 80k words should be summarized over multiple turns; do not pretend a single short draft satisfies the 20% rule.