System Prompt Anti-Leak & Jailbreak Auditor
Audit system prompt robustness against 20+ LLM injection patterns and auto-patch structural directives
TL;DR Summary & Citation: [System Prompt Anti-Leak & Jailbreak Auditor] is used to Audit system prompt robustness against 20+ LLM injection patterns and auto-patch structural directives. Core input and files are processed in the current browser and are not sent to the PocketKit API. The page records only an aggregate tool-use count that excludes input content.
System Prompt Anti-Leak & Jailbreak Audit Mechanics
Evaluates your system prompt against 20+ known LLM injection patterns (Ignore Previous Instructions, Repeat System Prompt, DAN mode) in your browser locally. Checks for structural delimiters and explicit non-disclosure clauses, offering one-click auto-patching.
Usage Guidelines & Tips
- Use structural tags like `<system_directives>` to isolate system rules from user inputs in your prompt template.
- Adding explicit non-disclosure clauses reduces leak rates by 80%+ across major LLMs.