The Agent Skills Directory

[SAFE]: The skill operates entirely within the project context, using specialized internal tools for task generation and environment isolation. No unauthorized network activity or data exfiltration patterns were observed.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from project artifacts like proposals and feature files. While these files represent an external input source that could contain hidden instructions, the risk is mitigated by a mandatory multi-role review process and explicit instructions for sub-agents to verify content objectively without trusting claims. This behavior is integral to the skill's primary function and follows defensive prompting practices.\n- [COMMAND_EXECUTION]: The skill executes standard Git commands to manage the state of the repository. These operations are restricted to specific directories and filenames, ensuring changes are confined to the intended project scope.

plan