Security category
Prompt Security
Prompt injection, instruction boundaries, untrusted input and output handling.
Prompt Security
Prompt Injection Indicator Analyzer
Identify suspicious prompt-injection indicators and unsafe trust-boundary patterns in supplied instructions.
Prompt SecurityAgent Instruction Security Analyzer
Inspect trusted instructions for unsafe interpolation of user or external content.
Prompt SecurityUntrusted Input Boundary Checker
Find indicators that external or untrusted content can flow directly into tools or trusted instructions.
Prompt SecurityAgent Output Handling Risk Checker
Detect unsafe execution, raw HTML rendering and query construction using model-generated output.