v1.3.3

What's Changed

• Added URL extraction from LLM instead of through docker for explore mode on VNC
• Added classification for new nodes added to existing graph
• Added safe capture screenshot functionality
• Enabled GPU acceleration
• Fixed images not updating for different domains in same route
• Fixed node image displaying incorrectly at times
• Fixed autolayout not initializing when adding new nodes
• Fixed invalid edge type issue
• Fixed regex incorrectly identifying emails as domain names
• Fixed context resetting randomly
• Updated factifai logo
• Improved graph data updates for newly available data
• Enhanced graph rendering and responsiveness
• Improved documentation generation chat bubble UI and prompt
• Added error handling for history parsing
• Enhanced UI elements in explore mode
• Improved prompts and added current URL integration

v1.3.0

What's Changed

• Explore mode UX refinements by @AzizStark in #12
• Explore mode and chat mode bug fixes by @AzizStark in #13
• Feat: Omni parser V2 integration by @sanju-presidio in #14

Features

• Integrated OmniParser v2: Delivers enriched, annotated screenshots to the LLM, enabling more informed decision-making for task execution
• Automated Playwright Installation: Playwright binaries now automatically install after npm install, streamlining the setup process.
• Factifai Logo Added: Improved visual identity with the addition of the Factifai logo.
• OpenAI Support for Explore Mode: Explore mode now supports OpenAI models, expanding LLM options and capabilities.
• Chat History and Persistence: Added chat history tracking with file storage persistence, allowing users to revisit previous conversations.

Bug Fixes

• LLM Context Isolation: Resolved context contamination between different operating modes, ensuring accurate and isolated responses.
• Chat Context Management: Implemented context management to prevent exceeding LLM token limits on complex websites.
• Explore Mode Graph Fix: Corrected a bug causing incorrect graph rendering in explore mode.
• Seamless VNC Mode Switching: Resolved issues with VNC mode switching, ensuring a smoother user experience.

Enhancements

• UX Improvements: General UX enhancements implemented to improve usability and overall user experience.

v1.2.0

What's Changed

• Explore Chat: Specialized chat interface for Click-through exploration of interconnected web content of a website. #11
• Graph View: Visual representation of web pages and their relationships for easier navigation and understanding of site structure #11
• Page Node System: Interactive page nodes that display content and allow navigation between related pages #11
• Recent Chats: Easy access to previous explore mode conversations #11

New Contributors

• @AzizStark #11

v1.1.0

Release Notes

Enhancements

• Added browser-centric approach on the puppeteer mode. (#9)
• General improvements & bugfixes (#9)

v1.0.0

Release Notes

Built-in support for leading vision-language models:

• Claude: Anthropic's advanced vision and reasoning model
• OpenAI: GPT-4o with visual understanding capabilities
• Gemini: Google's multimodal AI for computer interaction
• OmniParser: Screen Parsing tool for Pure Vision Based GUI Agent

AI-Powered Computer Control

• Intelligent element detection and navigation
• Automated verification and validation
• Comprehensive test documentation with automated screenshot capture for each step
• Integrated test case export with visual step-by-step documentation