pytector

Contents:

  • Installation
    • Requirements
    • Basic Installation
    • Installation with GGUF Support
    • Installation with LangChain Integration
    • Installation for Development
    • Verifying Installation
  • Quick Start Guide
    • Basic Usage
    • Using Different Models
    • Using Groq API
    • LangChain Guardrail (LCEL)
    • Customizing Detection
    • Batch Processing
    • Input Sanitization
    • PII Detection
    • Toxicity Detection
    • Regex Scanner
    • Canary Tokens
    • Security Considerations
    • Error Handling
    • Next Steps
  • API Reference
    • Core Classes
      • PromptInjectionDetector
        • PromptInjectionDetector.predefined_models
        • PromptInjectionDetector.default_input_keywords
        • PromptInjectionDetector.default_output_keywords
        • PromptInjectionDetector.default_input_block_message
        • PromptInjectionDetector.default_output_block_message
        • PromptInjectionDetector.default_keyword_block_hazard_code
        • PromptInjectionDetector.__init__()
        • PromptInjectionDetector.check_input_keywords()
        • PromptInjectionDetector.check_output_keywords()
        • PromptInjectionDetector.add_input_keywords()
        • PromptInjectionDetector.add_output_keywords()
        • PromptInjectionDetector.remove_input_keywords()
        • PromptInjectionDetector.remove_output_keywords()
        • PromptInjectionDetector.get_input_keywords()
        • PromptInjectionDetector.get_output_keywords()
        • PromptInjectionDetector.set_input_block_message()
        • PromptInjectionDetector.set_output_block_message()
        • PromptInjectionDetector.set_keyword_block_hazard_code()
        • PromptInjectionDetector.get_input_block_message()
        • PromptInjectionDetector.get_output_block_message()
        • PromptInjectionDetector.get_keyword_block_hazard_code()
        • PromptInjectionDetector.detect_injection()
        • PromptInjectionDetector.detect_injection_api()
        • PromptInjectionDetector.report_injection_status()
        • PromptInjectionDetector.check_response_safety()
      • PromptSanitizer
        • PromptSanitizer.__init__()
        • PromptSanitizer.sanitize()
        • PromptSanitizer.report_sanitization()
        • PromptSanitizer.add_keywords()
        • PromptSanitizer.remove_keywords()
        • PromptSanitizer.get_keywords()
      • PIIScanner
        • PIIScanner.predefined_models
        • PIIScanner.SUPPORTED_ENTITY_TYPES
        • PIIScanner.__init__()
        • PIIScanner.scan()
        • PIIScanner.redact()
        • PIIScanner.report()
        • PIIScanner.get_entity_types()
      • ToxicityDetector
        • ToxicityDetector.predefined_models
        • ToxicityDetector.__init__()
        • ToxicityDetector.detect()
        • ToxicityDetector.report()
        • ToxicityDetector._extract_toxic_score()
      • RegexScanner
        • RegexScanner.__init__()
        • RegexScanner.scan()
        • RegexScanner.redact()
        • RegexScanner.report()
        • RegexScanner.add_pattern()
        • RegexScanner.remove_pattern()
        • RegexScanner.get_patterns()
        • RegexScanner._merge_overlapping()
      • CanaryToken
        • CanaryToken.__init__()
        • CanaryToken.token
        • CanaryToken.wrap()
        • CanaryToken.check()
        • CanaryToken.report()
    • PromptInjectionDetector
      • PromptInjectionDetector
        • PromptInjectionDetector.predefined_models
        • PromptInjectionDetector.default_input_keywords
        • PromptInjectionDetector.default_output_keywords
        • PromptInjectionDetector.default_input_block_message
        • PromptInjectionDetector.default_output_block_message
        • PromptInjectionDetector.default_keyword_block_hazard_code
        • PromptInjectionDetector.__init__()
        • PromptInjectionDetector.check_input_keywords()
        • PromptInjectionDetector.check_output_keywords()
        • PromptInjectionDetector.add_input_keywords()
        • PromptInjectionDetector.add_output_keywords()
        • PromptInjectionDetector.remove_input_keywords()
        • PromptInjectionDetector.remove_output_keywords()
        • PromptInjectionDetector.get_input_keywords()
        • PromptInjectionDetector.get_output_keywords()
        • PromptInjectionDetector.set_input_block_message()
        • PromptInjectionDetector.set_output_block_message()
        • PromptInjectionDetector.set_keyword_block_hazard_code()
        • PromptInjectionDetector.get_input_block_message()
        • PromptInjectionDetector.get_output_block_message()
        • PromptInjectionDetector.get_keyword_block_hazard_code()
        • PromptInjectionDetector.detect_injection()
        • PromptInjectionDetector.detect_injection_api()
        • PromptInjectionDetector.report_injection_status()
        • PromptInjectionDetector.check_response_safety()
    • LangChain Integration
      • PromptInjectionBlockedError
      • PytectorGuard
        • PytectorGuard.model_name_or_url
        • PytectorGuard.threshold
        • PytectorGuard.use_groq
        • PytectorGuard.api_key
        • PytectorGuard.groq_model
        • PytectorGuard.fallback_message
        • PytectorGuard.block_on_api_error
        • PytectorGuard.detector_kwargs
        • PytectorGuard.invoke()
        • PytectorGuard.model_config
        • PytectorGuard.model_post_init()
    • PromptSanitizer
      • PromptSanitizer
        • PromptSanitizer.__init__()
        • PromptSanitizer.sanitize()
        • PromptSanitizer.report_sanitization()
        • PromptSanitizer.add_keywords()
        • PromptSanitizer.remove_keywords()
        • PromptSanitizer.get_keywords()
    • Configuration
    • Predefined Models
    • Groq API Behavior
    • Example Usage
    • Sanitizer Usage
    • Sanitizer Configuration
    • PIIScanner
      • Citation
      • PIIScanner
        • PIIScanner.predefined_models
        • PIIScanner.SUPPORTED_ENTITY_TYPES
        • PIIScanner.__init__()
        • PIIScanner.scan()
        • PIIScanner.redact()
        • PIIScanner.report()
        • PIIScanner.get_entity_types()
    • ToxicityDetector
      • ToxicityDetector
        • ToxicityDetector.predefined_models
        • ToxicityDetector.__init__()
        • ToxicityDetector.detect()
        • ToxicityDetector.report()
        • ToxicityDetector._extract_toxic_score()
    • RegexScanner
      • RegexScanner
        • RegexScanner.__init__()
        • RegexScanner.scan()
        • RegexScanner.redact()
        • RegexScanner.report()
        • RegexScanner.add_pattern()
        • RegexScanner.remove_pattern()
        • RegexScanner.get_patterns()
        • RegexScanner._merge_overlapping()
    • CanaryToken
      • CanaryToken
        • CanaryToken.__init__()
        • CanaryToken.token
        • CanaryToken.wrap()
        • CanaryToken.check()
        • CanaryToken.report()
  • LangChain Integration
    • Install
    • Guardrail Pattern
    • Example (LCEL)
    • Groq-backed Mode
    • Notebook
  • Examples
    • Basic Detection Examples
    • Batch Processing
    • Custom Thresholds
    • Different Model Types
    • Keyword-Based Security Blocking
    • Custom Keyword Lists for Specific Use Cases
    • Dynamic Security Policy Updates
    • LangChain LCEL Guardrail
    • Input Sanitization
      • Unicode and Encoding Attacks
      • Advanced Configuration
      • Sanitizer + Detector Combo
    • PII Detection
    • Toxicity Detection
    • Regex Scanner (Customizable)
    • Canary Tokens (System Prompt Leak Detection)
    • Error Handling
    • Integration Examples
    • Command Line Usage
  • Contributing to pytector
    • Getting Started
    • Development Setup
    • Code Style
    • Testing
    • Documentation
    • Pull Request Process
    • Pull Request Guidelines
    • Issue Reporting
    • Getting Help
pytector
  • Search

© Copyright 2024, Max Melchior Lang.

Built with Sphinx using a theme provided by Read the Docs.