claude-skill-bash
Fail
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
scripts/scaffold.shscript is vulnerable to command injection via the--dependenciesargument. The input is interpolated directly into asedcommand string without sanitization or escaping. This allows for execution of arbitrary commands, especially on systems using GNU sed with the 'e' flag, or through command injection if the input can break out of the sed command context.\n- [COMMAND_EXECUTION]: Thescripts/scaffold.shutility automatically applieschmod +xto generated script files. While this is part of its core functionality, it performs automated permission changes on the local filesystem.\n- [PROMPT_INJECTION]: The skill's primary function is to analyze and modify user-provided bash code, creating an inherent surface for indirect prompt injection if those scripts contain malicious instructions for the AI agent.
Recommendations
- AI detected serious security threats
Audit Metadata