The Agent Skills Directory

[SAFE]: The skill outlines a procedural workflow for iterative development and review. It follows best practices for state management by creating dedicated directories and using git for version control contexts.
[COMMAND_EXECUTION]: The skill uses local shell commands like git rev-parse and mkdir to initialize the environment. These are legitimate operations for the skill's stated purpose of managing project-specific build iterations.
[REMOTE_CODE_EXECUTION]: The instructions involve executing a 'Verification policy' which may include running user-defined tests or code. This is an intended feature of the 'builder' role and is triggered by user goals rather than external unverified sources.
[PROMPT_INJECTION]: The skill includes instructions to ensure the agent does not 'game' the system (e.g., not editing goals to make them easier). This acts as a safeguard for maintaining task integrity.

iterate