The Agent Skills Directory

[COMMAND_EXECUTION]: Utilizes the ! shell execution syntax to perform file system checks (ls, grep) within the docs/plans/ directory. These commands are used solely for prerequisite verification and do not expose sensitive data or perform network operations.
[SAFE]: Includes a dedicated post-implementation verification step that triggers both a code-verifier and a security-reviewer sub-agent, ensuring that the generated code aligns with design documents and security standards.
[SAFE]: The workflow requires explicit user approval for task decomposition and includes multiple escalation checkpoints if sub-agents encounter blocking issues or require revision, maintaining human-in-the-loop control.

recipe-front-build