The Agent Skills Directory

[SAFE]: The skill is designed for local code analysis and formatting using established tools (luacheck and stylua). It includes explicit instructions to avoid silent tool installation, ensuring user oversight.
[SAFE]: The skill contains no network-enabled commands (e.g., curl, wget) and explicitly states it does not collect or exfiltrate user code or credentials.
[SAFE]: No obfuscation, prompt injection, or privilege escalation patterns were detected in the instructions or examples.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted Lua files as part of its primary function. While this presents a theoretical attack surface, the risk is minimized by the use of standard static analysis tools and the lack of complex data processing or downstream execution of linter output. Evidence chain: 1. Ingestion point: Local .lua files. 2. Boundary markers: Standard tool output boundaries. 3. Capability inventory: Local subprocess execution of luacheck and stylua. 4. Sanitization: Handled by the underlying diagnostic tools.

codeguard-lua