apex
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads source code and project metadata to plan and implement features. This creates a surface for indirect prompt injection where malicious instructions embedded in project files could influence the agent's behavior.
- Ingestion points: File reading and codebase exploration during the 'Analyze', 'Plan', and 'Execute' steps (SKILL.md, references/workflow.md).
- Capability inventory: The skill has the ability to modify local files, execute build/test commands, and push code to remote repositories using git and gh.
- Boundary markers: The prompts for sub-agents do not include specific safety delimiters or 'ignore instructions' warnings for user-provided code content.
- Sanitization: No explicit sanitization of codebase content is performed before interpolation into sub-agent prompts.
- [COMMAND_EXECUTION]: The skill facilitates the execution of local development tools and project-specific scripts defined in project configuration files.
- Evidence: Automated detection and execution of build, lint, and test commands (e.g., npm run build, cargo test) identified in Steps 5 and 9.
- Evidence: Direct usage of git and GitHub CLI (gh) for repository management and Pull Request creation.
Audit Metadata