subagent-driven-development
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
delegate_tasktool withterminalandfiletoolsets to execute shell commands, perform tests, and modify source code during the implementation phase. - [INDIRECT_PROMPT_INJECTION]: The skill processes implementation plans from external files. As these files constitute untrusted data that guides the behavior of subagents with high-privilege tool access (filesystem and terminal), they present a surface for potential indirect prompt injection attacks.
Audit Metadata