v0.9.5

added up-to-date models for gemini, claude, openai

v0.9.4

Release Notes

🎉 New Features

LLM Enhancements

• Added GLM-5 Model Support - Extended language model capabilities with GLM-5 model integration for improved multilingual and reasoning tasks

Agent-to-Agent (A2A) Server

• Persistent Context Storage - Implemented Redis/persistent context store for A2A server enabling state persistence across sessions and improved reliability

v0.9.3

✨ Features

• LLM Model Support: Added support for Claude Opus 4.6 model (#4a842aa)

🔧 Enhancements

• Thinking/Reasoning: Extended thinking budget support for Opus and reasoning models, enabling advanced reasoning capabilities on additional model types (#c5fc99b)
• Docker: Minimized Docker dependencies to reduce image footprint and improve deployment efficiency (#a27544f)

🐛 Bug Fixes

• Thinking Configuration: Fixed issue where reasoning/thinking mode could not be disabled after being enabled (#cf7f35c)

📦 Dependencies

• Reduced Docker build dependencies for lighter container images

v0.9.2

hotfix: GUI cannot be loaded due to annotations

v0.9.1

🎉 New Features

Memory Management

• Delete conversation memory: Deleting a conversation now also removes associated memory entries (fba3bf5)

Console Improvements

• Visual mode for message selection: Added visual mode in console for selecting and copying messages (c98ec94)

File Operations

• Document file support: get_file now supports reading document files directly (PDF, DOCX, XLSX, PPTX, images) (3006d66)

LLM Models

• DeepInfra models update: Updated available models for DeepInfra integration (53a60e6)

---

🔧 Fixes

Model Support

• OSS model structured output: Added proper structured_output handling for OSS models (dff84be)

MCP Integration

• MCP stability: Fixed issue where MCP server would terminate on failed ping attempts (e58557619)

---

⚡ Performance & Maintenance

Compatibility

• A2A agent card compatibility: Updated agent card compatibility (2bf20c7)

---

All changes merged as of Feb 4, 2026

v0.9.0

🎉 Release Highlights

✨ Features

• Console Search: Added search functionality to conversation list for easier navigation
• Debug Tool: Revised /debug tool for improved readability and user experience
• Delete Feature: Implemented delete functionality in console conversation browser

🔧 Improvements & Refactoring

• GUI Theme System: Refactored GUI theme handler for better theme management and flexibility
• Main App: Refactored core application structure for improved maintainability
• Conversation Browser: Enhanced UI/UX with dynamic conversation list items
• Context Management: Added current working directory (cwd) to context for better file path handling
• Docker Configuration: Optimized Docker configuration by reusing common options

🐛 Bug Fixes

• Token Management: Fixed token limit calculation to use dynamic max context tokens from model instead of fixed 128k value
• GitHub Copilot: Corrected context window configuration for GitHub Copilot integration

📝 Other Changes

• Updated terminology: Changed "shrinked" wording for better clarity
• General conversation browser refactoring for improved code organization

Release Period: January 23 - January 27, 2026

v0.8.13

🎉 New Features

Conversation Browser for Console UI

• Added a new conversation browser interface for the console UI, improving user experience when reviewing conversation history and interactions.

Intelligent Code Analysis with Token Efficiency

• Implemented intelligent code analysis capabilities that reduce token consumption through hierarchical tree-based analysis, making the agent more cost-effective and responsive.

Improved Directory Awareness

• Enhanced the agent's understanding of directory structure and context, enabling better navigation and file management capabilities.

GLM-4.7 Models Integration

• Added support for GLM-4.7 models from DeepInfra, expanding available language model options for the agent.

Release Notes Display

• Print release notes before upgrade, keeping users informed about changes before they update.

---

🔧 Improvements & Changes

Command Execution Limit Increase

• Increased the command execution limit to allow for more complex workflows and longer-running operations.

File Editing Tool Refactoring

• Simplified file editing tool arguments for a clearer and more intuitive API
• Revised the file editing tool structure for improved understanding and usability

Code Analysis Tool Refactoring

• Refactored the code analysis tool for better maintainability and performance

---

🐛 Bug Fixes

Memory Management with Topics

• Fixed issues when the agent forgets memory with specific topics, ensuring better context retention during conversations.

C# Property Display

• Fixed display of both type and name for C# properties in code analysis.

---

📦 Other Changes

Funding Option

• Added funding option to support project development and maintenance.

---

Release Date: January 22, 2026

v0.8.12

Release Notes

🎯 New Features

• GrepTextService for Cross-Platform Text Searching (#37)

  • Added GrepTextService for robust, cross-platform text pattern searching
  • Integrated grep_text tool definition and handler for seamless LLM integration
  • Registered grep_text tool in main registration function
  • Comprehensive unit tests covering all aspects of the service

• Enhanced File and Function Analysis Limits

  • Limited analyzed files to 400 files maximum to prevent context overflow
  • Limited function/class output to prevent context overflow and improve performance

• Browser Interaction Improvements

  • Added refresh browser tool for improved navigation control
  • Fixed send key tool to work correctly across different scenarios

🔧 Improvements & Refactoring

• Adaptive Behavior Enhancements

  • Upgraded adaptive behavior for more complex instruction handling
  • Improved agent focus to prioritize task breakdown

• Memory Management Optimizations

  • Enhanced memory prompt for better context retention
  • Limited memory headers retrieval to last 20 entries for efficiency
  • Re-ordered user messages for improved message flow

• Tool Descriptions & UX

  • Refined tool descriptions for better utilization
  • Suppressed unnecessary exceptions inside httpcore when stopping streams

📝 Notes

All changes maintain backward compatibility while improving performance and user experience. The new text searching capabilities and file analysis limits significantly enhance the system's ability to handle larger codebases without context overflow issues.

v0.8.11

GitHub Release Changelog - v0.8.11

🎉 New Features

Code Analysis & File Search

• find_files tool (#36) - Added comprehensive file search functionality across platforms
  • New FileSearchService with cross-platform search capabilities and fallback mechanisms
  • LLM-compatible tool definition and handler for file search operations
  • Full integration with existing code analysis tools

Voice & LLM Enhancements

• Voice: Migrated to chatterbox-turbo for improved DeepInfra service integration
• LLM Models: Updated OSS models for better compatibility and performance
• LLM: Added Gemini 3 Flash support for faster inference

Agent & MCP Improvements

• Adaptive Instructions: Enhanced agent instruction system for better task-specific responses
• MCP Refresh Token Mechanism: Implemented automatic token refresh for Model Context Protocol
• Agent Capabilities: Agent can now decline tasks by default for improved safety and accuracy
• MCP SDK: Upgraded Model Context Protocol SDK to latest version

🐛 Bug Fixes & Improvements

Command Execution

• Ensured command execution runs in foreground mode for better control and visibility
• Agent now properly tracks background command execution status

File Editing & Syntax Detection

• Fixed syntax detection issues with C# language support
• Improved syntax check logging for better debugging

User Experience

• Minor console UI improvements for better readability
• Added effect delays between API calls for smoother user experience
• Refined user pre-messages for clearer communication

Performance Optimization

• Optimized consolidation logic to reduce token usage and improve efficiency
• Streamlined agent tool result handling

📦 Dependencies

• Upgraded MCP (Model Context Protocol) SDK

---

Changelog generated from commits between v0.8.10 and v0.8.11
Release Date: January 13, 2026

v0.8.10

✨ Features

• GPT-5.2 Model Support (commit 08dc409)

  • Added support for the latest GPT-5.2 model to the LLM module

• Automatic Content Shrinking on Token Limit (commit dca93df)

  • Implemented automatic content shrinking when API returns token exceeded error
  • Improves resilience when dealing with large content payloads

🔧 Improvements

• Performance Enhancement (commit 6608c6d)
  • Migrated to Python's built-in tomllib for TOML parsing
  • Improved performance and fixed several reading bugs
  • Better reliability when handling configuration files