* Update LLM providers and constants
- Updated constants in app/lib/.server/llm/constants.ts
- Modified stream-text functionality in app/lib/.server/llm/stream-text.ts
- Updated Anthropic provider in app/lib/modules/llm/providers/anthropic.ts
- Modified GitHub provider in app/lib/modules/llm/providers/github.ts
- Updated Google provider in app/lib/modules/llm/providers/google.ts
- Modified OpenAI provider in app/lib/modules/llm/providers/openai.ts
- Updated LLM types in app/lib/modules/llm/types.ts
- Modified API route in app/routes/api.llmcall.ts
* Fix maxCompletionTokens Implementation for All Providers
- Cohere: Added maxCompletionTokens: 4000 to all 10 static models
- DeepSeek: Added maxCompletionTokens: 8192 to all 3 static models
- Groq: Added maxCompletionTokens: 8192 to both static models
- Mistral: Added maxCompletionTokens: 8192 to all 9 static models
- Together: Added maxCompletionTokens: 8192 to both static models
- Groq: Fixed getDynamicModels to include maxCompletionTokens: 8192
- Together: Fixed getDynamicModels to include maxCompletionTokens: 8192
- OpenAI: Fixed getDynamicModels with proper logic for reasoning models (o1: 16384, o1-mini: 8192) and standard models
- Updated constants in app/lib/.server/llm/constants.ts
- Modified stream-text functionality in app/lib/.server/llm/stream-text.ts
- Updated Anthropic provider in app/lib/modules/llm/providers/anthropic.ts
- Modified GitHub provider in app/lib/modules/llm/providers/github.ts
- Updated Google provider in app/lib/modules/llm/providers/google.ts
- Modified OpenAI provider in app/lib/modules/llm/providers/openai.ts
- Updated LLM types in app/lib/modules/llm/types.ts
- Modified API route in app/routes/api.llmcall.ts
## Summary
Comprehensive fix for AI models (Claude 3.7, DeepSeek) that output code to chat instead of creating workspace files. The enhanced parser automatically detects and wraps code blocks in proper artifact tags.
## Key Improvements
### 1. Enhanced Message Parser
- Detects code blocks that should be files even without artifact tags
- Six pattern detection strategies for different code output formats
- Automatic file path extraction and normalization
- Language detection from file extensions
### 2. Pattern Detection
- File creation/modification mentions with code blocks
- Code blocks with filename comments
- File paths followed by code blocks
- "In <filename>" context patterns
- HTML/Component structure detection
- Package.json and config file detection
### 3. Intelligent Processing
- Prevents duplicate processing with block hashing
- Validates file paths before wrapping
- Preserves original content when invalid
- Automatic language detection for syntax highlighting
## Technical Implementation
The solution extends the existing StreamingMessageParser with enhanced detection:
- Falls back to normal parsing when artifacts are properly tagged
- Only applies enhanced detection when no artifacts found
- Maintains backward compatibility with existing models
## Testing
✅ Tested with various code output formats
✅ Handles multiple files in single message
✅ Preserves formatting and indentation
✅ Works with all file types and languages
✅ No performance impact on properly formatted messages
This fix ensures consistent file creation regardless of AI model variations.
🚀 Generated with human expertise
Co-Authored-By: Keoma Wright <founder@lovemedia.org.za>
When AI models fail to use proper artifact tags, code blocks now get
automatically detected and converted to file artifacts, preventing code
from appearing in chat. The parser detects markdown code fences outside
artifacts and wraps them with proper artifact/action tags.
This fixes the issue where code would randomly appear in chat instead
of being generated as files in the workspace.
Fixes#1230
Co-Authored-By: Keoma Wright <founder@lovemedia.org.za>
- Changed the maxTokenAllowed property to use Math.min for limiting the value to a maximum of 16384 tokens, ensuring better control over context window size.
- Introduced a `sanitizeText` function to remove specific HTML elements and content from messages, enhancing the integrity of the streamed text.
- Updated the `streamText` function to utilize `sanitizeText` for both user and assistant messages, ensuring consistent message formatting.
- Adjusted message processing to maintain the structure while applying sanitization.
add glowing effect component for tab tiles
improve tab tile appearance with new glow effect
add 'none' log level and simplify log level handling
simplify tab configuration store by removing developer tabs
remove useDebugStatus hook and related debug functionality
remove system info endpoints no longer needed
- Implemented a function to detect project frameworks based on package.json and configuration files.
- Added support for including source files in the deployment request for frameworks that require building.
- Updated the action function to handle framework detection and adjust deployment configuration accordingly.
- Removed unnecessary console logs and improved error handling for file reading operations.
Add MCP integration including:
- New MCP settings tab with server configuration
- Tool invocation UI components
- API endpoints for MCP management
- Integration with chat system for tool execution
- Example configurations
- Refactored the new fine-tuned system prompt to heavily reduce token usage by removing redundent snippets and duplicate instructions while keeping the same strict rules in place.
- Default prompt now uses this system prompt to reduce token usage by default
- Move user provided design scheme section to be grouped with other design instructions
- update user message icon color to use accent-500
- Enhance wording for better professionalism and clarity in design scheme usage
- Implement element inspector tool for preview iframe with hover/click detection
- Add inspector panel UI to display element details and styles
- Integrate selected elements into chat messages for reference
- Style improvements for chat messages and scroll behavior
- Add inspector script injection to preview iframe
- Support element selection and context in chat prompts
-Redesign Messgaes, Workbench and Header for a more refined look allowing more workspace in view
- Implement design scheme system with palette, typography, and feature customization
- Add color scheme dialog for user customization
- Update chat UI components to use design scheme values
- Improve header actions with consolidated deploy and export buttons
- Adjust layout spacing and styling across multiple components (chat, workbench etc...)
- Add model and provider info to chat messages
- Refactor workbench and sidebar components for better responsiveness
- Implement discuss mode toggle and UI in chat box
- Add quick action buttons for file, message, implement and link actions
- Extend markdown parser to handle quick action elements
- Update message components to support discuss mode and quick actions
- Add discuss prompt for technical consulting responses
- Refactor chat components to support new functionality
The changes introduce a new discuss mode that allows users to switch between code implementation and technical discussion modes. Quick action buttons provide immediate interaction options like opening files, sending messages, or switching modes.
Remove redundant checks for deleted paths and streamline binary file handling. This fixes the browser using excessive memory and freezing.
Improve DiffView to use a singleton instance of Shiki
- Added logging for dynamic max tokens based on model details.
- Increased max token limit for Claude model from 8000 to 128000.
- Included beta header for Anthropik API call.
Remove redundant error type handling in webcontainer to simplify logic and improve maintainability. Additionally, comment out lock file patterns to speed up npm install process.
* feat: Add reusable UI components and fix GitHub repository display
* style: Fix linting issues in UI components
* fix: Add close icon to GitHub Connection Required dialog
* fix: Add CloseButton component to fix white background issue in dialog close icons
* Fix close button styling in dialog components to address ghost white issue in dark mode
* fix: update icon color to tertiary for consistency
The icon color was changed from `text-bolt-elements-icon-info` to `text-bolt-elements-icon-tertiary`
* fix: improve repository selection dialog tab styling for dark mode
- Update tab menu styling to prevent white background in dark mode
- Use explicit color values for better dark/light mode compatibility
- Improve hover and active states for better visual hierarchy
- Remove unused Tabs imports
---------
Co-authored-by: KevIsDev <zennerd404@gmail.com>
* Add persistent file locking feature with enhanced UI
* Fix file locking to be scoped by chat ID
* Add folder locking functionality
* Update CHANGES.md to include folder locking functionality
* Add early detection of locked files/folders in user prompts
* Improve locked files detection with smarter pattern matching and prevent AI from attempting to modify locked files
* Add detection for unlocked files to allow AI to continue with modifications in the same chat session
* Implement dialog-based Lock Manager with improved styling for dark/light modes
* Add remaining files for file locking implementation
* refactor(lock-manager): simplify lock management UI and remove scoped lock options
Consolidate lock management UI by removing scoped lock options and integrating LockManager directly into the EditorPanel. Simplify the lock management interface by removing the dialog and replacing it with a tab-based view. This improves maintainability and user experience by reducing complexity and streamlining the lock management process.
Change Lock & Unlock action to use toast instead of alert.
Remove LockManagerDialog as it is now tab based.
* Optimize file locking mechanism for better performance
- Add in-memory caching to reduce localStorage reads
- Implement debounced localStorage writes
- Use Map data structures for faster lookups
- Add batch operations for locking/unlocking multiple items
- Reduce polling frequency and add event-based updates
- Add performance monitoring and cross-tab synchronization
* refactor(file-locking): simplify file locking mechanism and remove scoped locks
This commit removes the scoped locking feature and simplifies the file locking mechanism. The `LockMode` type and related logic have been removed, and all locks are now treated as full locks. The `isLocked` property has been standardized across the codebase, replacing the previous `locked` and `lockMode` properties. Additionally, the `useLockedFilesChecker` hook and `LockAlert` component have been removed as they are no longer needed with the simplified locking system.
This gives the LLM a clear understanding of locked files and strict instructions not to make any changes to these files
* refactor: remove debug console.log statements
---------
Co-authored-by: KevIsDev <zennerd404@gmail.com>
- Add support for `PREVIEW_CONSOLE_ERROR` in WebContainer error handling
- Introduce new Search component for text search functionality
- Extend `ScrollPosition` interface to include `line` and `column`
- Implement scroll-to-line functionality in CodeMirrorEditor
- Add tab-based navigation for files and search in EditorPanel
This commit introduces several enhancements to the editor, including improved error handling, better scrolling capabilities, and a new search feature. The changes are focused on improving the user experience and adding new functionality to the editor components.
Clean up code by removing unnecessary debug logging statements in `StarterTemplates.tsx` and `useShortcuts.ts`. Making it easier to debug issues in console
Refine the UI design and content guidelines in the prompts to ensure consistency and professionalism. Add detailed instructions for animations, color schemes, typography, and layout.
Remove redundant console log in the LLM stream-text module.
Remove unused imports, files parameter, and redundant multimodal handling logic. Streamline the function by directly passing processed messages to _streamText. Also, add specific handling to remove package-lock.json content to reduce token usage
The dependency array in the Artifact component was missing `artifact.type` and `allActionFinished`, which could lead to incorrect evaluation of action completion. Additionally, the logic for determining if all actions are finished was updated to account for 'start' actions that are 'running'. This ensures that the component accurately reflects the state of bundled artifacts.
- Restructured alert components in BaseChat for better layout organization
- Updated artifact component to display dynamic titles based on state
- Simplified template names in constants for better readability
- Enhanced snapshot restoration process by consolidating command actions into a single artifact
Improve code maintainability by relocating the ScrollToBottom function outside the BaseChat component. This reduces complexity and enhances readability.
Introduce a new fine-tuned prompt for better results and update mobile app development instructions to ensure comprehensive guidance. The changes include enhanced design guidelines, improved database handling, and clearer artifact creation rules for better project setup.
- Updated package.json and pnpm-lock.yaml to use react-qrcode-logo v3.0.0
- Modified ExpoQrModal.tsx to use the new QRCode component with enhanced styling and logo support
- Removed filtering of lock.json files in useChatHistory.ts and stream-text.ts for consistency
- Updated mobile app instructions in prompts.ts to ensure clarity and alignment with best practices
Update the design instructions to emphasize the importance of content richness and realistic placeholders. This ensures applications feel functional and visually appealing immediately, avoiding generic templates and empty screens.
To improve data consistency and reliability, snapshot storage has been migrated from localStorage to IndexedDB. This change includes adding a new 'snapshots' object store, updating database version to 2, and modifying related functions to use IndexedDB for snapshot operations. The migration ensures better handling of snapshots alongside chat data and removes dependency on localStorage preventing UI lag.
Refine mobile app development instructions to ensure clarity and consistency. Enhance styling guidelines with detailed design system requirements, including color, typography, and responsive design. Update critical requirements for components, animations, and error handling to align with best practices.
The useSnapScroll hook has been replaced with the StickToBottom component to improve the scrolling behavior in the chat interface. This change ensures smoother and more consistent scrolling, especially when new messages are added. The StickToBottom component provides better control over the scroll position and handles edge cases more effectively.
This commit introduces the `refreshAllPreviews` method in the `PreviewsStore` class, which iterates through all previews and triggers a file change broadcast for each. This ensures that all previews are updated after a file save operation.
refactor(CodeBlock): handle unsupported languages by falling back to plaintext
The `CodeBlock` component now defaults to 'plaintext' when an unsupported language is detected, improving the user experience by avoiding unsupported language errors.
prompts: update dependency installation instructions
The prompts documentation has been updated to clarify the process of installing dependencies, emphasizing the importance of updating `package.json` first and avoiding individual package installations.
Refactor and consolidate mobile app development and design guidelines in the prompts files. Remove redundant information and ensure clarity and consistency in the instructions. The changes aim to provide a more structured and concise set of guidelines for developers working on mobile apps using Expo and React Native.
- Clarify critical instructions regarding artifact creation and image files
- Add detailed design instructions for visual identity, UX, and layout
- Include guidance on using Unsplash for stock photos and realistic placeholder content
Refactor the file deletion logic in FilesStore to precompute prefixes and iterate through files only once. This reduces the complexity of nested loops and improves performance by applying all deletions in a single update to the store. Additionally, remove a redundant console.log statement in Chat.client.tsx and update the prompts documentation for clarity.
Add 'grok-3-beta' to xAI provider and 'gemini-2.5-flash-preview-04-17' to Google provider. Also, ensure file saving when content is updated in WorkbenchStore and update streaming indicator styling in chat messages.
The qrCodeStore has been relocated from the app/stores directory to app/lib/stores to maintain a more consistent and organized project structure. This change improves maintainability by centralizing store-related files in a dedicated directory.
Remove the unused `qrCodeAtom` from the QR code store to clean up the codebase. Additionally, update the mobile app development prompts to emphasize critical requirements, such as creating a default route, ensuring high-quality UI/UX, and following Expo best practices.
