The Agent Skills Directory

[SAFE] (SAFE): No security issues detected across the 10 threat categories.
[PROMPT_INJECTION] (SAFE): No override markers, role-play injections, or system prompt extraction attempts found in GUIDE.md or other files.
[DATA_EXFILTRATION] (SAFE): No sensitive file paths or network operations identified; code only accesses local skill assets.
[COMMAND_EXECUTION] (SAFE): Scripts use standard libraries for file validation without spawning risky subprocesses or executing shell commands.
[EXTERNAL_DOWNLOADS] (SAFE): No remote code execution patterns or unverifiable package installations detected.
[CREDENTIALS_UNSAFE] (SAFE): No hardcoded API keys, tokens, or secrets found in assets/config.yaml or script files.

machine-learning