react-patterns
Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill identifies user-provided code as the primary input and explicitly instructs the agent to treat this content as data rather than instructions. This acts as a mitigation against indirect prompt injection where a user might include instructions inside the code to be reviewed.
- [DATA_EXFILTRATION]: Analysis of the instructions and examples confirms there are no network operations, credential harvesting attempts, or accesses to sensitive local filesystem paths.
- [COMMAND_EXECUTION]: The skill does not contain any patterns for executing shell commands, system scripts, or other process-level operations. It is confined to generating textual feedback and code corrections.
- [EXTERNAL_DOWNLOADS]: No remote dependencies or external code fetches are utilized in the skill's logic or prerequisites.
Audit Metadata