Skip to content

GuardScan v1.0.5

Latest
Latest

Choose a tag to compare

View all tags
@github-actions github-actions released this 09 Dec 19:10
· 1 commit to main since this release
v1.0.5
2ba2a61

[1.0.5] - 2025-12-09

Added

AI Model Selection

  • Model Selection in Configuration: Added ability to select specific AI models during guardscan config:
    • OpenAI: Choose from gpt-5.1, gpt-4o, gpt-4.1-mini, gpt-3.5-turbo
    • Claude: Choose from claude-opus-4.5, claude-sonnet-4.5, claude-haiku-4.5
    • Gemini: Choose from gemini-3-pro, gemini-2.5-pro, gemini-2.5-flash, gemini-2.5-flash-lite
  • Model Override: Added --model flag to override AI model for individual commands

Enhanced Chat Experience

  • Improved Chat Output Formatting: Enhanced guardscan chat with rich terminal formatting:
    • Colored file paths (cyan) for better visibility
    • Highlighted inline code snippets (yellow)
    • Formatted code blocks with language labels and borders
    • Bold text styling for emphasis
    • Word wrapping for better readability
    • Structured statistics display with clear formatting
  • Chat Export Functionality: Added /export command to save conversations:
    • Exports to markdown format with full conversation history
    • Automatically saves to parent directory with timestamped filename
    • Includes session metadata and relevant files referenced
    • Format: guardscan-chat-{sessionId}-{timestamp}.md
  • Interactive Chat Commands: Enhanced chat interface with commands:
    • /help - Display available commands and example questions
    • /clear - Clear conversation history
    • /stats - Show session statistics (messages, tokens, duration)
    • /export - Export conversation to markdown
    • /exit or /quit - Exit chat session

Documentation

  • Comprehensive Chat Guide: Added detailed docs/CHAT_GUIDE.md covering:
    • RAG (Retrieval-Augmented Generation) explanation
    • Interactive commands and usage
    • CLI options and customization
    • Example use cases and best practices
    • Troubleshooting guide
    • Privacy and security information
  • Updated Getting Started Guide: Enhanced docs/GETTING_STARTED.md with interactive chat section
  • Updated Quick Start: Added chat commands and export functionality to QUICKSTART.md
  • Updated README: Added references to chat features and new documentation

Changed

AI Provider Updates

  • Updated OpenAI Models:
    • Added gpt-5.1, gpt-4o, gpt-4.1-mini
    • Removed deprecated models (gpt-4-turbo-preview, gpt-4-vision-preview, gpt-4-32k)
    • Updated default model to gpt-4o
    • Updated pricing for current models
  • Updated Claude Models:
    • Added claude-opus-4.5, claude-sonnet-4.5, claude-haiku-4.5
    • Removed deprecated models (claude-3-opus, claude-3-sonnet, claude-3-haiku, claude-2.1)
    • Updated default model to claude-sonnet-4.5
    • Updated pricing for current models
  • Updated Gemini Models:
    • Added gemini-3-pro, gemini-2.5-pro, gemini-2.5-flash, gemini-2.5-flash-lite
    • Removed deprecated models (gemini-1.5-pro, gemini-1.5-flash, gemini-pro)
    • Updated testConnection() to try multiple models in priority order
    • Enhanced error handling with specific error messages for API key validation
    • Updated pricing for current models

Configuration Improvements

  • Enhanced Config Display: Improved guardscan config output with:
    • Numbered section headings for better organization
    • Detailed descriptions for each configuration option
    • Clear status indicators for enabled/disabled features
    • Better visual hierarchy and readability
  • Better Error Messages: Enhanced error reporting for AI provider connection issues:
    • Specific error messages for invalid API keys
    • Model availability checking with fallback logic
    • Clear instructions for troubleshooting

CI/CD Improvements

  • Simplified NPM Publishing: Updated GitHub Actions workflow:
    • Publish job now only runs on version tags (not on every main branch push)
    • Removed unnecessary dry-run step
    • Simplified conditional logic for better maintainability
  • Updated Node.js Testing: Changed test matrix to only test on Node.js 20 (removed Node.js 18)

Fixed

Test Suite

  • Fixed Embedding Search Tests: Resolved TypeScript errors in embedding-search.test.ts:
    • Added missing loadIndex() method to MockEmbeddingStore
    • Added missing checkCompatibility() method to MockEmbeddingStore
    • Updated saveEmbeddings() signature to match interface
  • Fixed Empty Store Handling: Improved embedding-search.ts to properly handle empty embedding stores:
    • Returns empty results instead of throwing error for empty stores
    • Only throws error when embeddings exist but have incompatible dimensions
    • Better error messages for dimension mismatch scenarios

Embedding Provider Factory

  • Fixed TypeScript Compilation: Resolved type error in embedding-factory.ts:
    • Fixed embeddingFallback type checking logic
    • Improved fallback selection for Claude provider

Chat Engine

  • Fixed Model Parameter Passing: Ensured model selection is properly passed through the entire call chain:
    • Updated ChatOptions to include model parameter
    • Modified callAI() to pass model to AI provider
    • Updated all command files to pass config.model to provider factory
    • Fixed assistantMsg.metadata.modelUsed to use actual model from API response

Technical Improvements

Code Quality

  • All 327 tests passing across 25 test suites
  • Improved type safety with proper TypeScript interfaces
  • Enhanced error handling throughout the codebase
  • Better separation of concerns in provider implementations

Architecture

  • Consistent model parameter handling across all AI providers
  • Improved factory pattern for provider instantiation
  • Better metadata tracking for AI responses
  • Enhanced session management for chat feature
Assets 2
Loading