plan-technical-review
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill defines a structured workflow for reviewing implementation plans using internal agents and templates. It utilizes standard tool patterns like
AskUserQuestionand file system operations within thedocs/plan/directory. - [PROMPT_INJECTION]: The skill processes untrusted content from user-provided implementation plan files, creating a surface for indirect prompt injection.
- Ingestion points: Implementation plan files passed as the
path to plan fileargument (SKILL.md). - Boundary markers: None explicitly defined to separate untrusted plan content from agent instructions.
- Capability inventory: File writing (
docs/plan/), agent invocation (@code-simplicity-review-agent, etc.), and tool execution (AskUserQuestion). - Sanitization: No explicit sanitization or validation of the input plan content is performed before processing.
Audit Metadata