Modify selectFile function to check for existing tabs before creating new ones.
When clicking file paths in AIChat, it will now switch to the existing tab
instead of creating duplicates.
- Detect file paths with dots in directory names (e.g. next/styles/SignIn.module.css)
- Create new files when path ends with "(new file)"
- Use existing socket connection and file management system
### Summary
- Added a new "Apply" button to code snippets provided by the AI assistant.
- The button is designed to seamlessly merge the AI-generated snippet into the relevant file in the editor.
### Current Issues
1. **Sticky Accept/Decline Buttons:** These activate for every snippet instead of being limited to the relevant snippet.
2. **Discard Button:** Currently non-functional.
3. **Highlight Inconsistencies:** The green-red code highlights for old and new code are inconsistent.
### To Do
- Implement a toast notification when the "Apply" button is pressed on an irrelevant tab to prevent code application errors.
### Workflow Implemented
1. The "Apply" button is added alongside "Copy" and "Reply" for AI-generated code snippets.
2. Upon clicking "Apply," the code snippet and relevant file content (active file) are sent to a secondary model (GPT-4O).
3. The system prompt for GPT-4O instructs it to merge the snippet with the file content:
- Ensure the original file functionality remains intact.
- Integrate the code snippet seamlessly.
4. The output from GPT-4O is injected directly into the code editor.
5. Changes are visually highlighted:
- Green for new code.
- Red for removed code.
6. Highlights remain until the user explicitly accepts or discards the changes.
- Add template configurations with file structures and conventions
- Update AI route handler to include template context in system messages
- Pass template type through AIChat component
- Add template-specific run commands
- Enhance AI responses with project structure knowledge
- Move hardcoded run commands from navbar/run.tsx to templates/index.ts
This improves the AI's understanding of different project templates (React, Next.js, Streamlit, Vanilla JS) and enables more contextual assistance based on the project type.
- Remove separate AI worker service
- Added generation limits:
FREE: 1000/month (For the beta version)
PRO: 500/month
ENTERPRISE: 1000/month
- Integrate AI functionality into main API routes
- Added monthly generations reset and usage tier upgrade API routes
- Upgrade tier page to be added along with profile page section
- Created new config file for ignored paths in file system traversal
- Separated ignored folders and files into dedicated arrays
- Includes comprehensive ignore patterns for:
- Package managers (node_modules, venv)
- Build outputs and caches
- Version control
- IDE specific folders
- Framework specific directories
- System and config files
- Lock files and compiled assets
- added context tabs
- added multifile context including file and image uploads to the context along with all the files from the project
- added file/image previews on input
- added code paste from the editor and file lines recognition
- added image paste from clipboard and preview
- cmd/ctrl + L works globally now
- added the copilot and ai chat button indicators
- when aichat is open, the preview/terminal column becomes horizontal
features:
1. Real-time message display
2. User input handling
3. AI response generation
4. Markdown rendering for AI responses
5. Syntax highlighting for code blocks
6. Copy to clipboard functionality for messages and code blocks
7. Context handling (setting, displaying, and removing context)
8. Expandable/collapsible context display
9. Ability to ask about specific code snippets
10. Auto-scrolling to the latest message
11. Loading indicator during AI response generation
12. Stop generation functionality
13. Error handling for failed API requests
14. Responsive design (flex layout)
15. Custom styling for user and AI messages
16. Support for various Markdown elements (paragraphs, lists, code blocks)
17. Language detection and display for code blocks
18. Animated text generation effect for AI responses
19. Input field placeholder changes based on context presence
20. Disable input during message generation
21. Send message on Enter key press
22. Expandable/collapsible message context for each message
23. Editable context in expanded view
24. Icons for various actions (send, stop, copy, expand/collapse)
25. Visual feedback for copied text (checkmark icon)
26. Abortable fetch requests for AI responses
27. Custom button components
28. Custom loading dots component
29. Truncated display of long messages with expand/collapse functionality
Akhileshrangani4