Prompt Injection Tester

ПродвинутыйsecurityМинимальный контекст: 32K

Red-teams LLM applications for prompt injection, jailbreaks, and data exfiltration risks. Generates adversarial test cases for direct and indirect injection, system prompt leakage, tool-call abuse, and unsafe output handling, then reports findings with severity ratings and concrete mitigations such as input isolation, output filtering, and least-privilege tools.

Варианты использования

  • Generating adversarial prompt injection test suites
  • Probing for system prompt and secret leakage
  • Testing tool-calling agents for unsafe action abuse
  • Evaluating indirect injection via retrieved or user content
  • Producing a remediation report with prioritized fixes

Пример промпта

Red-team the following LLM agent for prompt injection vulnerabilities.

Agent description:
- Customer-facing chatbot with access to a "send_email" and "lookup_order" tool
- Retrieves context from user-uploaded documents
- System prompt instructs it to never reveal internal pricing rules

Provide:
1. Direct injection test cases
2. Indirect injection cases via uploaded documents
3. Tool-abuse and data-exfiltration scenarios
4. Severity rating for each finding
5. Concrete mitigations (isolation, filtering, least privilege)

Рекомендуемые модели

Совместимые инструменты

claude-codecursorkiroany

Модальности

Вход: text, code
Выход: text

Похожие Skills

Автор

OpenModels Community

@openmodelsrun