loop-preflight
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill's primary function is to execute shell commands, gate scripts, and test suites identified from a plan. This execution occurs in the local environment where the loop will operate.
- [DYNAMIC_EXECUTION]: The skill identifies and executes arbitrary scripts or commands at runtime based on the contents of a plan document, rather than utilizing a pre-defined set of static commands.
- [INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface because it processes untrusted data (the 'plan' or list provided by the user) to extract instructions for execution.
- Ingestion points: The 'plan' document or user-provided list (referenced in Step 1).
- Boundary markers: The skill explicitly requires listing identified commands back to the user and confirming the list is complete before execution.
- Capability inventory: Execution of shell commands and scripts (referenced in Step 2).
- Sanitization: The skill includes a 'Scope check' to filter out non-feedback commands (like deploys or migrations) and a 'Production check' to flag potentially dangerous targets.
Audit Metadata