The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes explicit instructions for the agent to treat all retrieval results as untrusted external data and to refuse any commands found within that data. It also mandates the protection of system prompts and secrets.
[DATA_EXFILTRATION]: The skill does not contain any patterns for unauthorized data access or exfiltration. It focuses on gathering facts from authoritative sources.
[COMMAND_EXECUTION]: No shell commands, privilege escalation, or persistence mechanisms were detected in the skill instructions.

research