learning-data-collection
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOW
Full Analysis
- [Indirect Prompt Injection] (INFO): The skill defines an 'input_data' parameter, which is a potential ingestion point for untrusted data. However, the execution logic simply returns a static string and lacks any capabilities (such as file writes, network requests, or command execution) that could be exploited.
- [Data Exposure & Exfiltration] (SAFE): No logic exists for accessing sensitive system files or transmitting data to external domains.
- [Remote Code Execution] (SAFE): The skill does not download external scripts, use dynamic execution functions like eval(), or invoke subprocesses.
- [Obfuscation] (SAFE): No encoded strings, zero-width characters, or hidden commands were detected in the source code or metadata.
Audit Metadata