The Agent Skills Directory

[SAFE]: The skill serves as a high-level engineering specification and identity guide. All analyzed files focus on promoting architectural integrity, safety invariants, and surgical communication protocols.\n- [SAFE]: Data exposure and exfiltration risks are mitigated by explicit directives. The 'Principal Dossier' (V16, S11, S17) and 'Anti-Patterns' (T0.2) specifically prohibit writing secrets to logs and emphasize least privilege.\n- [SAFE]: No remote code execution or unverifiable dependency patterns were found. The 'Toolchain Registry' provides architectural recommendations (e.g., Zod, Prisma, PostgreSQL) but does not automate installations or execute remote scripts.\n- [SAFE]: No prompt injection or behavior bypass patterns were detected. While the skill establishes a 'Rule of Precedence' for its engineering principles, these rules are designed to enhance safety and code quality rather than bypass platform constraints.\n- [SAFE]: No obfuscation, persistence mechanisms, or privilege escalation techniques were identified within the documentation or metadata.

posture-core