Releases: owulveryck/gomcptest
v1.5.1
Changelog
- cbbe30d Fix artifact storage upload routing issue causing POST to GET redirect
Released by GoReleaser.
v1.5.0
Changelog
- 1cc603b Add WebM audio file support to AgentFlow UI
- 56e8ba4 Add clipboard paste support for images in AgentFlow UI
- 646a2cb Add comprehensive AgentFlow explanation documentation following Divio framework
- 3b159a0 Add comprehensive audio file support to Gemini implementation and AgentFlow UI
- b60bb8b Adjust AgentFlow UI code block font size for better readability
- 93d062b Document -withAllEvents flag across README and docs for mandatory tool event testing
- 6dfbde6 Documentation update
- e21fca0 Enhance AgentFlow UI with advanced code blocks and file attachment features
- 99ef56e Enhance documentation with granular implementation details and practical tool querying guide
- 34c74ac Fix AgentFlow UI copy button to preserve markdown formatting
- 1772db3 Fix Google Search tool reactivation bug in stream processing follow-ups
- 1802952 Fix Vertex AI Google Search tool activation bug in FilterTools method
- 3d33cf0 Fix audio display sizing and improve usability
- 9618cd6 Fix system audio recording compatibility issues
- fcd2cef Housekeeping
- f320a20 Implement Google search tool exclusivity in AgentFlow UI
- f4708e0 Implement artifact storage API with comprehensive documentation
- 00b1a33 Implement artifact storage for large audio recordings in AgentFlow UI
- 09e0c2a Implement comprehensive error event system with red UI styling
- c95225f Implement comprehensive in-app audio recording feature for AgentFlow UI
- a187d80 Implement conversation export/import functionality in AgentFlow UI
- c12ac5b Implement default tool unchecking behavior in AgentFlow UI
- 3267c79 Improve audio attachment display with better icons and sizing
- aebfeda Improve documentation to follow Diátaxis framework principles
- 08c0df8 Optimize audio wave icon size and container height for better UX
- 6585ff1 Reduce code block font size in AgentFlow UI for improved compactness
- 2960483 Refactor UI to use embedded AgentFlow with improved serving
- 14c3462 Replace segment button icon with more meaningful split/divide icon
- d83a4c3 Unify AgentFlow attachment buttons into single universal button
- 32f7332 Update documentation for v1.4.0 - introduce AgentFlow UI and new environment variables
Released by GoReleaser.
v1.4.0
Changelog
- e9ab533 Add PDF upload support to OpenAI server and AgentFlow UI
- c6a479c Add PlantUML MCP tool with server and testing infrastructure
- f7ece76 Add a template engine for the prompt system as middleware
- 3f2b35d Add pipe-separated model format and tool selection UI
- f965eee Add support for ExecutableCode and CodeExecutionResult parts in streaming
- 1cd7e4b Better version
- e0d27f5 Deleted binaries
- 871659f Enhance AgentFlow UI with professional styling, accessibility, and conversation management
- 2c4c708 Fix MCP tool addition to prevent tool type mixing
- f6bfb6c Fix PlantUML integration and UI edit button errors
- 2961fc7 Fix localStorage quota exceeded and CORS issues in AgentFlow UI
- b40a35d Fix message editing with attachments and add attachment removal functionality
- 2dcc079 Fix multimodal content handling in conversation title generation
- b5da249 Fix template
- b38474e Fix tool filtering to prevent mixing tool types
- 379dae5 Fixed error
- 8004244 Ignoring the MCP tests
- 0b67b94 Improve message selection mapping in chat UI
- 3ad9da5 Improve unsupported part type error message with detailed information
- ace802a Refactor PlantUML tool to URL-based architecture with decoding support
- 0976563 Replace PNG favicon with SVG favicon
- 2a87523 feat: Add complete Apple touch icon support for iPhone web app
- 0176b1a feat: Add multiple Apple touch icon sizes for iPhone web app
- 6da97e5 feat: Optimize chat UI for iPhone and mobile devices
- 54e55b0 fix: Correct MCP server argument parsing for single arguments
Released by GoReleaser.
v1.3.1
v1.3.0
Changelog
- ce9e441 feat: Add clickable tool notifications with popup details and proper alignment
- c7636ce feat: Add persistent tool notification in conversation flow
- 96ab723 fix: Correct tool notification ordering to appear before assistant response
- 9d71068 fix: Improve copy button positioning and selection behavior
- aa97ef6 fix: Improve markdown copy functionality and button positioning
- fb19e2a fix: Improve tool popup positioning and auto-close timing
- ad7b855 fix: Resolve assistant response display issue after tool notifications
- d85f319 fix: Save assistant messages in streaming responses and adjust popup timeout
Released by GoReleaser.
v1.2.0
Changes from v1.1 to current version
✨ Major New Features
🎉 Complete Chat UI Implementation
- Brand New Chat Interface: Added comprehensive web-based chat UI for the OpenAI server
- Embedded in Server: Chat UI accessible at
/uiendpoint, fully integrated with the OpenAI server - Professional Design: Modern, responsive interface with markdown support and syntax highlighting
- Real-time Streaming: Live response streaming with visual indicators and typing animations
- Conversation Management: Full conversation system with sidebar, auto-titling, rename/delete, and localStorage persistence
- System Prompt Support: Customizable system prompts per conversation with save/reset functionality
- Message Editing: Edit any message and regenerate responses from that point
- Dynamic Model Selection: Choose AI models directly from the interface
- Tool Execution Visibility: Real-time display of tool calls and responses with detailed popups
- SVG Rendering: Full support for visualizations including Wardley maps and diagrams
Enhanced Tool Integration
- Sleep MCP Tool: Added new sleep tool for debugging and testing UI interactions
- Streaming Events: Comprehensive streaming event system showing tool execution in real-time
- Tool Call Correlation: Proper matching of tool calls with their responses using unique IDs
🔧 Technical Improvements
Streaming Architecture
- Event-Driven Streaming: Refactored chat engine with StreamEvent interface for unified event handling
- Gemini Integration: Enhanced Gemini provider with tool call and response events
- Provider Switch: Changed default provider from Claude to Gemini
- Enhanced Flags: Added
--withAllEventsflag for comprehensive event streaming
Server Architecture
- Improved Routing: Updated to use
http.ServeMuxfor better endpoint management - CORS Handling: Proper cross-origin request support for the chat UI
- Resource Management: Better cleanup and error handling throughout the system
🎨 User Experience Features
Interface Design
- Responsive Layout: Optimized for different screen sizes with minimal margins
- Professional Theme: Clean color scheme maximizing usable space
- Sliding Sidebar: Hover-activated conversation management with arrow indicator
- Visual Feedback: Success/error indicators, loading states, and progress animations
- Accessibility: Proper contrast, readable fonts, and intuitive navigation
Interactive Features
- Conversation Persistence: All conversations saved locally across browser sessions
- Auto-titling: Conversations automatically named from first message
- Typing Indicators: Visual feedback while AI is processing requests
- Tool Popups: Detailed execution status and results in overlay windows
- Message History: Complete conversation replay and editing capabilities
🐛 Bug Fixes
SVG Display
- Object Tag Rendering: Fixed SVG display using
<object>tags for better browser compatibility - Responsive Sizing: Improved SVG sizing to use full available space with proper aspect ratios
- Marked.js Compatibility: Fixed object parameter handling in custom image renderer
Tool Popup System
- Popup Persistence: Resolved issues with popups getting stuck or not closing properly
- Correlation Fixes: Ensured tool calls properly match with their responses
- Timeout Handling: Added appropriate timeouts and fallbacks for unresponsive tools
- Visual Improvements: Enhanced popup size, readability, and formatting
📋 Development Notes
This release represents a major milestone with the addition of a complete web-based chat interface. The UI provides a professional, feature-rich environment for interacting with the MCP system and demonstrates the full capabilities of the agentic framework.
The chat UI includes advanced features like real-time tool execution visibility, conversation management, and comprehensive streaming support, making it a powerful tool for testing and using MCP-based agents
v1.1
Summary
Version 1.1 represents a significant enhancement to the gomcptest project, focusing on improved Google Gemini integration, comprehensive error handling, and expanded MCP tool capabilities.
🚀 New Features
Enhanced Google Gemini Integration
- Upgraded Dependencies: Updated
google.golang.org/genaifrom v1.16.0 to v1.22.0 - Function Calling Configuration: Added FunctionCallingConfig with validated mode for better tool handling
- Tool Configuration: Implemented ToolConfig with FunctionCallingConfigModeValidated for improved reliability
New MCP Tools
- PlantUML Check Tool: Added comprehensive PlantUML syntax validation tool
- Enhanced Imagen Edit Tool: Expanded image editing capabilities with HTTP server functionality
Advanced Error Handling
- UNEXPECTED_TOOL_CALL Handling: Added specific error handling for unexpected tool call scenarios
- Detailed Error Messages: Implemented comprehensive error reporting with troubleshooting suggestions
- Stream Processing Improvements: Enhanced error recovery in both streaming and non-streaming chat processors
🔧 Improvements
Logging Enhancements
- Content-Based Logging: Improved logging to show actual content instead of memory pointers
- Structured Error Reporting: Added detailed context for debugging tool call failures
- Better Debug Information: Enhanced logging with model versions, response IDs, and usage metadata
Development Workflow
- Claude Code Permissions: Updated settings to allow additional git commands and file access
- Build System: Enhanced Makefile for better tool management
- Resource Management: Improved handling of external dependencies
Error Recovery
- MCP Server Failures: Comprehensive error handling for MCP server connection failures
- Tool Call Validation: Better validation and error recovery for malformed tool calls
- Safety Ratings: Enhanced handling of content filtering and safety issues
📁 File Changes
Core Components Modified (7 files changed, 147 insertions, 12 deletions)
go.mod&go.sum: Dependency updateshost/openaiserver/chatengine/gcp/: Complete overhaul of GCP chat enginenonstream.go: Enhanced non-streaming request handlingnonstream_processor.go: Improved response processing with error handlingstream.go: Updated streaming functionalitystream_processor.go: Comprehensive error handling and tool call management
New Tool Additions
tools/imagen_edit/: Complete image editing tool suite (830+ lines)tools/plantuml_check/: PlantUML validation tool (131+ lines)
Configuration Updates
.claude/settings.local.json: Extended permissions for development workflow
🐛 Bug Fixes
- Compilation Errors: Fixed various compilation issues across the codebase
- Tool Call Failures: Improved handling of malformed or invalid tool calls
- Stream Processing: Enhanced stability in streaming chat responses
- Resource Cleanup: Better management of external resources and connections
📊 Statistics
- Total Files Changed: 64
- Lines Added: 6,214
- Lines Removed: 25
- New Tools: 2 (imagen_edit, plantuml_check)
- Core Commits: 11 commits between versions
🔄 Migration Notes
For Developers
- Update your
go.modto use the new genai version (v1.22.0) - The new function calling configuration provides better tool validation
- Enhanced error messages will provide more debugging context
For Users
- Improved stability when using tools with the chat engine
- Better error reporting when tool calls fail
- New image editing and PlantUML validation capabilities
🎯 Key Commits
- 81066cf: feat: Update Google Gemini integration and dependencies
- b7571b9: feat: Add comprehensive error handling and new MCP tools
- 5ed8692: Implement comprehensive error handling for MCP server failures
- 3d76792: Improve logging to show actual content instead of memory pointers
- b36a132: Fix compilation errors and improve error handling across the codebase
🚀 What's Next
This release establishes a more robust foundation for:
- Advanced agent interactions with better error recovery
- Enhanced tool ecosystem with validation capabilities
- Improved developer experience with better debugging information
- Preparation for future MCP protocol enhancements
v1.0.0
Changelog
- bc23a5a Add comprehensive Imagen tool suite with HTTP server
- e75dee8 Add resource
- 7f7e446 Merge pull request #24 from owulveryck/better-logs
- aad2870 Merge pull request #25 from owulveryck/updated_packages
- de15e7b Merge pull request #26 from owulveryck/updated_packages
- 0d1ce7e Merge remote-tracking branch 'origin/resources'
- 8756ce1 README and CLAUDE update
- 33b6eb3 Rationalize Makefiles by consolidating into single root Makefile
- 7ab0614 Remove experimental imagen feature from host implementations
- 6f59016 Update Hugo documentation to reflect current codebase
- ef70570 Update architecture diagram to reflect current tool suite
- ed09e25 Update documentation to reflect current codebase state
- 0d781ce Update documentation to remove imagen environment variables
- 735c75d feat
- c31c667 feat: better loggin mechanism
- 9a3ca76 fix
- ae91f74 fix: in error message, do not try to cast the interface
- 61d51a2 resources
- 069de74 to be tested
- b5c53b7 updated everything
- f6e0709 updated version
Released by GoReleaser.