The Agent Skills Directory

[PROMPT_INJECTION]: No attempts to override agent behavior, bypass safety filters, or extract system instructions were detected. The instructions follow standard routing and documentation patterns.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths (like .ssh or .env), or unauthorized data exfiltration patterns were found. The skill operates within the scope of graphics programming and system diagnostics.
[OBFUSCATION]: The content is provided in clear text. There is no evidence of Base64 encoding for commands, zero-width characters, homoglyphs, or other obfuscation techniques.
[REMOTE_CODE_EXECUTION]: No remote script execution (e.g., curl|bash) or installation of unverifiable packages from untrusted sources was identified. Mentions of command-line tools like metal-shaderconverter, dxc, and metalperftrace are standard for the described developer workflows.
[COMMAND_EXECUTION]: Shell commands provided for metalperftrace and usdcrush are used for local diagnostic data collection and asset optimization, which is consistent with the primary purpose of a graphics development skill.
[DYNAMIC_CONTEXT_INJECTION]: The skill does not use the !command syntax to execute shell commands at load time.

axiom-graphics