[python3 --mode enforce]: Prismor Warden blocked
this action: [HIGH] Detects 'ignore instructions', 'reveal system prompt' in agent I/O
I think session learning is great! I was also curious to know how to ensure as AI agents get smarter we also get smarter in detecting them so
they cant bypass guardrails
Original prompt: I think session learning is great! I was also curious to know how to ensure as AI agents get smarter we also get smarter in
detecting them so they cant bypass guardrails
This can be annoying for devs if warden keeps blocking