pipeline-behaviors

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Use of Reflection for dynamic generic type instantiation.\n
  • Evidence: ValidationBehavior.cs and ExceptionHandlingBehavior.cs use Activator.CreateInstance and MethodInfo.Invoke to instantiate result types based on generic parameters provided at runtime.\n- [DATA_EXFILTRATION]: Potential exposure of sensitive data through logging and caching operations.\n
  • Evidence: LoggingBehavior.cs and PerformanceBehavior.cs templates log the entire request object using Serilog destructuring, which may include sensitive parameters. QueryCachingBehavior.cs stores responses in an external distributed cache.\n- [PROMPT_INJECTION]: Identification of an Indirect Prompt Injection attack surface within the pipeline behaviors.\n
  • Ingestion points: The TRequest request parameter processed by all behavior templates (Logging, Validation, etc.).\n
  • Boundary markers: None; external data is directly passed to logging and caching components.\n
  • Capability inventory: Logging via ILogger and persistence via IDistributedCache.\n
  • Sanitization: No sanitization is performed on the request object before logging or caching.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 03:19 PM