The Agent Skills Directory

[PROMPT_INJECTION]: No attempts to override agent behavior, bypass safety filters, or extract system prompts were detected.
[DATA_EXFILTRATION]: No network operations, sensitive file paths, or hardcoded credentials were found. The instructions explicitly advise avoiding sending sensitive data to external models without approval.
[REMOTE_CODE_EXECUTION]: The skill consists entirely of markdown instructions and does not perform any remote code downloads or package installations.
[COMMAND_EXECUTION]: No shell commands, subprocess spawning, or dynamic context injection patterns (e.g., backtick commands) are present.
[OBFUSCATION]: The content is clear-text markdown with no signs of Base64, hex encoding, zero-width characters, or homoglyph attacks.
[INDIRECT_PROMPT_INJECTION]: While the skill describes capabilities for building RAG systems and agents that process external data, the skill itself does not implement vulnerable data ingestion or prompt interpolation logic.

ai-engineer