The Agent Skills Directory

[SAFE]: The skill's primary functional component is scripts/slop_check.py, a Python script used to analyze text for 'AI slop'. The script uses only Python standard libraries (argparse, json, re, statistics, sys, zipfile, dataclasses).
[SAFE]: Analysis of scripts/slop_check.py shows no network operations (no urllib, requests, or socket usage), no shell command execution (no subprocess or os.system), and no dynamic code execution (no eval or exec).
[SAFE]: The script reads local files (Markdown, LaTeX, .txt, and .docx) to perform regex-based text analysis. The handling of .docx files via the zipfile module is a standard and safe implementation for extracting XML content from Word documents.
[DATA_EXPOSURE_AND_EXFILTRATION]: No evidence of hardcoded credentials, sensitive file path access, or network exfiltration was found. The script only outputs analysis results to the terminal or a local file if redirected by the user.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to process untrusted user-provided academic drafts, it lacks the necessary capabilities (network access, file-writing, or shell execution) to be exploited via indirect prompt injection. The analysis is purely static and regex-based.
[DYNAMIC_CONTEXT_INJECTION]: No use of the !command syntax or other dynamic injection patterns was detected in the markdown files.
[OBFUSCATION]: No obfuscated strings, Base64-encoded commands, homoglyphs, or hidden characters were found across any of the files.

unslop-academic-writing