The Agent Skills Directory

[PROMPT_INJECTION]: No evidence of instructions designed to bypass safety filters or override core agent behavior.
[DATA_EXFILTRATION]: No network operations, hardcoded credentials, or sensitive file path references (e.g., .ssh, .env) were detected.
[REMOTE_CODE_EXECUTION]: The skill does not include package installations or mechanisms to download and execute external scripts.
[OBFUSCATION]: Analysis found no hidden content, Base64-encoded commands, homoglyphs, or zero-width character techniques.
[PRIVILEGE_ESCALATION]: No commands for acquiring administrative rights or modifying system-level configurations are present.
[INDIRECT_PROMPT_INJECTION]: Although the skill involves reasoning about code snippets, it lacks the functional capabilities (file writing, shell access, or network requests) required to execute malicious instructions found in untrusted data.
[DYNAMIC_CONTEXT_INJECTION]: The skill does not utilize dynamic command execution syntax in its markdown instructions.

javascript-runtime