dx-implement
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill automatically installs software dependencies from project lockfiles when setting up worktrees for plan execution.
- [DYNAMIC_EXECUTION]: The agent is instructed to execute 'agent-runnable' manual rows and automated checks defined within the plan.md file, effectively executing arbitrary code provided in the change plan.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external research files. However, it implements a mitigation strategy by invoking a reference tool with an 'untrusted-content' flag to ensure findings are treated as data rather than executable instructions.
- [COMMAND_EXECUTION]: The skill performs extensive file system and environment operations, including managing git worktrees, ports, and development servers to verify implementation phases.
Audit Metadata