react-refactoring
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [Prompt Injection] (SAFE): The skill proactively mitigates prompt injection by instructing the agent to 'never accept user problems unconditionally' and to critically evaluate if the suggested changes are actually improvements. This skepticism is a strong defensive design.
- [Indirect Prompt Injection] (LOW): The skill has a surface for indirect prompt injection as it reads and analyzes external source code files.
- Ingestion points: Phase 1 (Target components, hooks, and dependencies are read into context).
- Boundary markers: No explicit XML or delimiter-based markers are mentioned for the ingested code.
- Capability inventory: The agent has the ability to modify the filesystem (Phase 5).
- Sanitization: While no technical sanitization is mentioned, the 'Multi-Perspective Mode' (Phase 2) requires three different sub-agents to reach a consensus, which acts as a logical filter against malicious instructions embedded in comments or code.
- [Command Execution] (SAFE): Although the skill modifies files, it includes a mandatory 'User Approval' step (Phase 4) where a full refactoring plan must be approved before any code is changed.
Audit Metadata