setup-build-tools

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION] (HIGH): The skill instructs the agent to execute a shell script (install-build-tools.sh) located in the project's hidden hooks directory to install system packages.
  • Evidence: bash "$CLAUDE_PROJECT_DIR/.claude/hooks/install-build-tools.sh" in SKILL.md.
  • Risk: Installing system-level tools (dpkg-dev, nodejs, imagemagick) typically requires root/sudo privileges. Executing scripts from the project directory allows for arbitrary code execution if the project files are untrusted.
  • [PROMPT_INJECTION] (HIGH): The skill incorporates a raw argument placeholder ($ARGUMENTS) at the end of the execution block without any sanitization or boundary markers.
  • Ingestion points: $ARGUMENTS placeholder in SKILL.md.
  • Boundary markers: None; external input is interpolated directly into the instruction context.
  • Capability inventory: Shell command execution (bash) and system package management.
  • Sanitization: None provided; the agent is not instructed to validate or escape the contents of $ARGUMENTS.
  • Risk: An attacker could provide malicious input via $ARGUMENTS to inject additional shell commands (e.g., using && or ;) or to override the agent's primary mission via indirect prompt injection.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 12:32 PM