design-cli-interface
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/check_cli.pyexecutes shell commands using thesubprocessmodule to test CLI behavior, such as flag handling and exit codes. While the instructions warn the user to only run trusted code, this provides a mechanism for arbitrary command execution. - [INDIRECT_PROMPT_INJECTION]: The skill processes and analyzes the output of external command-line tools, which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: Captures
stdoutandstderrfrom commands executed inscripts/check_cli.py. - Boundary markers: The captured tool output is evaluated without specific markers or safety instructions to distinguish it from the agent's internal logic.
- Capability inventory: The skill possesses shell execution capabilities through the
check_cli.pyscript. - Sanitization: Output from tested commands is not sanitized before being returned to the agent for analysis.
Audit Metadata