The Agent Skills Directory

[SAFE]: The skill does not contain any prompt injection attempts or instructions meant to bypass safety guardrails. All instructions are focused on task decomposition and planning logic.
[SAFE]: Data access is limited to reading the local repository for context and writing the resulting plan to a local directory (e.g., .plan/). No network exfiltration, hardcoded credentials, or unauthorized file access patterns were detected.
[SAFE]: No remote code execution, dynamic script generation, or untrusted package installations are present. The skill relies on platform-native interaction and standard file system operations.

closed-plan