Skills
skills/usorama/rad-engineer-v2/quality-gate/Gen Agent Trust Hub

quality-gate

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill possesses a high-risk attack surface by executing untrusted content from the project repository.\n
  • Ingestion points: Repository scripts referenced via bun run and the hook script located at .claude/hooks/check-system-resources.sh.\n
  • Boundary markers: None identified. The skill implicitly trusts the repository structure and script definitions.\n
  • Capability inventory: Can execute arbitrary shell commands via the Bash tool, including sh -c and bun subcommands.\n
  • Sanitization: No verification of the integrity or content of project scripts before execution.\n- [Command Execution] (MEDIUM): Uses shell tools to interact with system state, specifically sysctl and ps, and manages process locks via flock in /tmp. While intended for stability, these provide system visibility and interaction.\n- [Remote Code Execution] (MEDIUM): The reliance on a local hook script (.claude/hooks/check-system-resources.sh) allows for arbitrary code execution if an attacker can commit files to the repository or influence the build environment.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 05:21 AM