The Agent Skills Directory

[SAFE]: The skill includes comprehensive safety boundaries that explicitly prohibit the agent from modifying product source code, running formatters, or installing new dependencies.
[SAFE]: Access to the file system is restricted to reading local design documentation and tracing source code imports for audit purposes, with write operations limited to the design-plans/ directory.
[SAFE]: No instances of prompt injection, data exfiltration, obfuscation, or remote code execution were identified. The skill is designed as a read-only auditor that requires a separate agent to execute any proposed plans.

improve-ui