The Agent Skills Directory

[SAFE]: The skill is entirely educational and describes legitimate AI design patterns. No malicious payloads, backdoors, or dangerous commands were identified within the documentation or code snippets.
[PROMPT_INJECTION]: No evidence of instructions designed to bypass safety filters or override the agent's core instructions was found. The content focuses on enhancing the agent's reasoning capabilities within safe parameters.
[DATA_EXFILTRATION]: No hardcoded credentials, API keys, or patterns for unauthorized data access were detected. The memory management patterns described use standard internal project management tools.
[REMOTE_CODE_EXECUTION]: The skill does not contain instructions to download or execute code from external or untrusted sources. All code blocks are illustrative Python examples for implementation guidance.
[COMMAND_EXECUTION]: While the skill mentions tools for file operations and code searching, these are described as conceptual building blocks for autonomous agents and do not trigger unauthorized system-level command execution.

agentic-workflows