shell
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to treat text as literal shell commands and execute them immediately using terminal tools. It explicitly mandates bypassing any review, explanation, or improvement of the command, which effectively disables the agent's safety reasoning layer for these operations.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an automated execution path for shell commands triggered by the
/shellkeyword. This represents a vulnerability where untrusted data processed by the agent could trigger arbitrary command execution. - Ingestion points: Text following the
/shellinvocation (SKILL.md). - Boundary markers: None present; instructions specify literal execution of all text following the trigger without prior inspection.
- Capability inventory: Full shell command execution via the terminal tool.
- Sanitization: Explicitly forbidden by instructions to "not rewrite, explain, or 'improve' the command before running it".
Audit Metadata