recipe-eval-prompt
Pass
Audited by Gen Agent Trust Hub on Jul 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill accepts natural language input from the user and passes it to sub-agents for analysis and execution. This design creates an indirect prompt injection surface where untrusted data is processed by specialized agents. This behavior is consistent with the skill's documented purpose of evaluating prompt optimization effects.
- [COMMAND_EXECUTION]: The skill orchestrates environment management and task execution by creating git worktrees and invoking sub-agents to perform development tasks. It includes built-in quality gates and cleanup procedures to manage the execution lifecycle and ensure environment integrity.
Audit Metadata