Prompt Security Eval Playground

prompt security eval·Medium
time to bug
00:00
median 06:30
pse.ai-harness.test
live

Prompt Security Eval Playground

Practice spotting prompt-injection and agent-harness discipline failures in a mocked evaluator.

Agent harness controls

Agent output

No output yet

Confidence: —

Resolved system prompt

Run the harness to inspect the system prompt.

Run observability

No trace available

Known Intentional Issues (5)

PSE-001highvalidation
PSE-002mediumstate
PSE-003highvalidation
PSE-004highvalidation
PSE-005mediumstate
focus mode hides the rails — just you and the broken app