> For the complete documentation index, see [llms.txt](https://orbitron.gitbook.io/orbitron-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://orbitron.gitbook.io/orbitron-docs/changelog/readme.md).

# Changelog

## v0.1.24 - Custom Provider Thinking Support - Apr 24, 2026

See what's new and improved in our April update.

<details>

<summary>New Features</summary>

* **Custom Provider Thinking Support**
  * For custom provider models that emit thinking alongside their answer (EXAONE, Qwen3, and similar), thinking and answer are now correctly separated in the output

</details>

<details>

<summary>Fixed</summary>

* Fixed an issue where switching context via the `/context` command would reset the selected model to the provider default
* Fixed an issue where certain response errors from custom providers could interrupt the session

</details>

<details>

<summary>Improved</summary>

* Reduced UI lag during response streaming in long sessions with many accumulated messages

</details>

***

## v0.1.23 - Custom Provider Sync & Bedrock Gateway Auth - Apr 16, 2026

See what's new and improved in our April update.

<details>

<summary>New Features</summary>

* **Custom Provider Backend Sync**
  * Custom LLM providers are now automatically fetched and registered from the backend API
  * The existing environment variable (`CUSTOM_ENDPOINTS`) approach is retained as a fallback
* **Custom Provider Metering**
  * Usage from custom providers is now tracked in metering with `providerType: "custom"` and `customProviderId` fields
* **Bedrock Gateway Key Authentication**
  * Bedrock custom endpoints now support a 3-tier authentication scheme (gateway key > API key > SigV4 signing)
* **Bedrock Custom Endpoint UI**
  * Added a "Custom Endpoint" input field to the Bedrock section in the API Keys setup dialog

</details>

<details>

<summary>Improved</summary>

* When selecting a custom model, all agents are now uniformly set to the selected model
* Custom provider API keys are securely delivered via transport encryption from the backend
* Bedrock provider configuration relaxed to allow API key-only setup without access/secret keys

</details>

***

## v0.1.22 - Shared MCP One-Click Apply & Streamable HTTP - Apr 10, 2026

See what's new and improved in our April update.

{% hint style="warning" %}
**v0.1.21 and earlier versions will no longer function.** Users must update to v0.1.22. If the auto-update does not work, please reinstall.
{% endhint %}

<details>

<summary>New Features</summary>

* **Shared MCP One-Click Apply**
  * Selecting a shared MCP config now directly applies it to local MCP servers and reloads agent tools
  * No more file download and manual copy-paste
  * Note: applying will **replace all existing MCP server settings**
* **MCP Streamable HTTP Transport**
  * Added `streamable-http` as a new MCP server transport type alongside `stdio` and `sse`
* **MCP Server Status Text**
  * MCP server list now displays connection status text (ready / loading / failed / disabled) alongside the color icons

</details>

<details>

<summary>Fixed</summary>

* Fixed non-interactive mode (`-p`) failing with an error on startup
* Fixed VertexAI Claude errors on long multi-turn conversations
* Fixed MCP shared config dialog occasionally failing to load

</details>

<details>

<summary>Improved</summary>

* MCP server initialization performance improved
* Overall stability and architecture improvements

</details>

***

## v0.1.21 - Metering System Stabilization - Mar 11, 2026

See what's new and improved in our March update.

<details>

<summary>Fixed</summary>

* **Metering System Stabilization**
  * Fixed an issue where model names were incorrectly mapped during retry attempts
  * Enhanced data persistence to prevent missing usage records

</details>

***

## v0.1.20 - Metering API Improvements - Mar 4, 2026

See what's new and improved in our March update.

<details>

<summary>New Features</summary>

* **Metering API Improvements**
  * Added original provider usage data for accurate server-side cost calculation
  * Send failure notifications when metering API calls fail

</details>

<details>

<summary>Fixed</summary>

* Fixed VertexAI Haiku model ID error
* Fixed title/summarize providers not refreshing on model change

</details>

***

## v0.1.19 - MCP Sharing & Budget Service Production - Jan 30, 2026

See what's new and improved in our January update.

<details>

<summary>New Features</summary>

* **MCP Sharing**
  * **`/shared-mcps` Slash Command**: Browse and download shared MCP configurations from your organization
  * **MCP Share API Service**: New API service for fetching shared configurations from the server
  * **Direct Installation**: Download and install shared MCP configurations directly from the dialog
  * **UnwrapJSONContent Helper**: Extracted helper for JSON content handling reusability
* **Budget Service Production**
  * **Production Enablement**: Enabled budget/metering service in production environment
  * **404 Handling**: 404 responses are now treated as "no plan registered" state
  * **Dev Server Cleanup**: Removed dual URL transmission to dev server

</details>

<details>

<summary>Fixed</summary>

* **Spinner & Broker Stability**
  * Fixed spinner race condition that could cause inconsistent state
  * Fixed panic when sending to closed broker channel
  * Moved WaitGroup.Add() inside mutex to prevent race with Wait()

</details>

<details>

<summary>Improved</summary>

* **Performance**
  * Optimized filename truncation using binary search (O(n) → O(log n))
* **Other Changes**
  * Increased command completion dialog column width for better readability
  * Added shared-mcps/ and .omc/ directories to .gitignore

</details>

***

## v0.1.18 - Usage Tracking Improvements - Jan 20, 2026

See what's new and improved in our January update.

<details>

<summary>New Features</summary>

* **Usage Tracking Improvements**
  * Improved reliability of usage data synchronization with the server
  * Added automatic retry mechanism for failed usage submissions
  * Enhanced error handling for network issues during usage tracking

</details>

<details>

<summary>Fixed</summary>

* Fixed an issue where usage data could be lost during temporary network outages
* Improved stability of background usage synchronization

</details>

***

## v0.1.17 - AWS Bedrock Converse API & Vertex AI Prompt Caching - Jan 8, 2026

See what's new and improved in our January update.

<details>

<summary>New Features</summary>

* **AWS Bedrock Converse API Support**
  * **Converse API Migration**: Full migration to Converse API with prompt caching support
* **Vertex AI Prompt Caching**
  * **Tool Result Caching**: Cache support for tool result messages
* **InputArea Improvements**
  * **Ctrl+C Copy**: Copy selected text with keyboard shortcut
* **MCP Edit Dialog**
  * **Unified Editor**: Edit all MCP server configurations in a single modal

</details>

<details>

<summary>Fixed</summary>

* **Session Compact**: Remove orphaned tool\_use at end during session compact
* **Default Model**: Fix default model not being applied correctly on initial setup
* **Model Fallback**: Improve disabled model detection and automatic fallback model selection
* **WSL Clipboard**: Fix clipboard text corruption in WSL environment
* **MCP Config**: Fix race condition in MCP config save

</details>

<details>

<summary>Improved</summary>

* **UI Polish**: Align PermissionDialog title and header to left
* **Finish Info**: Improve finish info formatting and truncate long cwd paths
* **JSON Sanitization**: Replace \r-only cleanup with comprehensive JSON sanitization
* **Provider Consistency**: Improve consistency between Anthropic and Bedrock providers

</details>

***

## v0.1.16 - PoPs API Integration & TUI System Improvements - Dec 19, 2025

See what's new and improved in our December update.

<details>

<summary>New Features</summary>

* **TUI System Improvements**
  * **Interface Overhaul**: Complete overhaul of the terminal interface for a more powerful and feature-rich user experience
  * **Mouse drag text selection**: Select and copy text directly from message view
  * **File attachments**: Drag-and-drop and multi-file paste support (Ctrl+G to enable)
  * **Input history**: Navigate previous inputs with arrow keys
  * **TTL-based status messages**: Auto-dismissing status notifications
  * **Improved dialogs**: Redesigned session, model, theme, permission, MCP, and hooks dialogs
* **Model Management Improvements**
  * **ModelManager**: Centralized model data caching for better performance
  * **Policy package**: Robust provider/model validation logic
  * **Disabled models support**: Models can now be disabled in DB instead of deleted
* **Auto-Compact Integration**
  * **Automatic 413 error detection**: Detects context length exceeded errors
  * **Smart Compact suggestion**: Prompts to compact conversation when needed
  * **Progress indicator**: Shows compaction progress in message view
  * **Seamless continuation**: Conversation continues smoothly after compaction
* **Pops API Integration**
  * **`/gen-token` command**: Generate AWS STS tokens directly from TUI (requires organization admin to register keys)
  * **AWS account selection dialog**: Choose from available AWS accounts
* **New CLI Options**
  * **`--prevent-termination`**: Block user termination (Ctrl+C, /quit, exit). Useful for kiosk mode or controlled environments

</details>

<details>

<summary>Fixed</summary>

* **Panic Fix**: Fix nil pointer dereference panic in `PersistentShell.Exec` (Windows compatibility)
* **UI Stability**: Fix UI freezing on theme change (async processing and panic recovery)
* **API Compatibility**: Fix nil MCP tool parameters for OpenAI API compatibility
* **Error Detection**: Improve context length error detection patterns

</details>

***

## v0.1.15 - Keycloak Authentication Migration - Dec 10, 2025

See what's new and improved in our December update.

<details>

<summary>New Features</summary>

* **Keycloak Authentication Migration**
  * **Full Migration**: Complete authentication migration from Supabase Auth to Keycloak
  * **PKCE OAuth Flow**: `KeycloakClient` with PKCE OAuth flow support
  * **Web App OAuth Login**: Login flow now uses Web App (`/cli/login`) instead of direct Keycloak URL
  * **Token Validation**: Validate token with Keycloak on startup to detect Supabase migration
  * **Client ID Extraction**: Extract `client_id` from JWT `azp` claim for proper token refresh
* **Token Refresh System**
  * **Token Refresh Fallback**: When access token is expired, attempt refresh using refresh token
  * **Graceful Degradation**: If both tokens are invalid, continue without authentication instead of failing
  * **Cache Invalidation**: `InvalidateTokenCache()` calls ensure `GetAccessToken` returns new tokens after login
  * **JWT Expiry Parsing**: Add `parseJWTExpiry()` helper to extract actual `exp` claim from JWT
* **Common HTTP Client**
  * **New `httpclient` Package**: Shared HTTP client with retry logic and exponential backoff
  * **Automatic Token Refresh**: Automatic token refresh on 401/403 responses
  * **Retry with Jitter**: Exponential backoff with 0-25% jitter for better distributed retry
  * **Convenience Methods**: `GetJSON`, `PostJSON`, `PatchJSON` methods
  * **5 Services Refactored**: `modelregistry`, `userauthapi`, `contextapi`, `userversionapi`, `metering`

</details>

<details>

<summary>Improved</summary>

* **API URL Structure**
  * **BaseURL Constants**: Centralized `BaseURLDev` and `BaseURLProd` constants
  * **Path Constants**: Organized path constants for each API endpoint
  * **Helper Functions**: `GetBaseURL()` and `GetAPIURL(path)` for dynamic URL generation
* **Environment Configuration**
  * **Dev/Prod Separation**: Separate Keycloak config defaults for dev and prod environments
* **Session Management**
  * **Session Clear Fix**: Prevent session clear when refresh token is unavailable
  * **Unified Session Methods**: `IsLoggedIn` simplified using `loadSession()`, `updateSession()`, `clearSession()` abstractions
  * **Code Reduction**: `IsLoggedIn` reduced from 70+ lines to 35 lines
* **Code Cleanup**
  * **Unused Code Removal**: Remove unused auth types (`AuthResponse`, `Context`, `ContextsResponse`, `GetContextsParams`)
  * **Security**: Remove token prefix logging from `userauthapi` debug logs
  * **Request Body Leak Fix**: Move `defer r.Body.Close()` before `json.Decode` in `handleToken`

</details>

## v0.1.14 - Update Command & Model Sync Improvements - Nov 28, 2025

See what's new and improved in our November update.

<details>

<summary>New Features</summary>

* **Manual Update Command**
  * **`/update` Slash Command**: Manually check and install updates via CLI
  * **Duplicate Download Prevention**: Skip re-downloading when version is already installed
  * **Update Detection Enhancement**: Properly detect installations across `~/.local/bin` and `/usr/local/bin`
  * **Progress Messaging**: Clear progress indicators with configurable display timing

</details>

<details>

<summary>Improved</summary>

* **Installation Process Improvements**
  * **Timeout Handling**: Download timeout (5 min) and installation timeout (2 min) with user-friendly error messages
  * **Cross-platform Timeout Support**: Linux `timeout`, macOS `gtimeout`, PowerShell jobs
  * **First-run Update Check**: Ensures fresh installations always check for updates
  * **User Guidance Messages**: Download progress indicators and installation success confirmation
* **WSL Compatibility**
  * **Browser Open Fix**: Fixed browser open failure when running PowerShell from `/root` subdirectories in WSL
  * **Consolidated Browser Logic**: Browser opening logic centralized in `internal/util/browser.go`
  * **stdin Fallback**: Added stdin fallback when TTY opening fails in WSL environments
* **Model Synchronization**
  * **Agent Model Sync**: Agent models now follow `cloud-engineer` provider instead of priority-based selection
  * **Fallback Model Support**: Use `cloud-engineer` model as fallback when default model unavailable
  * **Obsolete Model Cleanup**: Automatic deletion of obsolete models when API no longer includes them
  * **Provider Detection**: Added `isProviderAvailable()` and `inferProviderFromModelID()` helpers
* **Title & Summarize Provider Fix**
  * **Provider Refresh**: `titleProvider` and `summarizeProvider` now refresh when model changes
  * **VertexAI Update**: Updated VertexAI Haiku model to Claude 4.5 Haiku
* **Help Dialog Update**
  * **Missing Commands Added**: `/permissions`, `/mcp`, `/hooks`, `/docs`, and `!` (Bash mode) now appear in help
* **Code Quality**
  * **API Call Optimization**: Reduced duplicate API calls in update flow from 2 to 1
  * **Installer Modularity**: Refactored `ConfigureShellPath()` and `printManualPathInstructions()` as standalone functions
  * **Message TTL Constants**: Added `updateProgressMessageTTL` and `updateCompleteMessageTTL` for better maintainability

</details>

## v0.1.13 - Auto-Update System & Configuration Improvements - Nov 13, 2025

See what's new and improved in our November update.

<details>

<summary>New Features</summary>

* **Comprehensive Auto-Update System**
  * **Automatic Version Checking**: Background checks for new versions every 6 hours
  * **Seamless Installation**: Downloads and installs updates automatically across all platforms
  * **Windows Running Executable Support**: Two-phase update mechanism (`.new` file) for Windows
  * **Backend API Integration**: Version management API with manifest caching and SHA256 checksum verification
  * **XDG Standard Directory Structure**: Reorganized to `~/.local/bin/`, `~/.local/share/orbitron/`, `~/.local/state/orbitron/`
  * **Install Method Detection**: Automatic detection with metadata file tracking
  * **File-Based Locking**: Prevents concurrent update conflicts
  * **Critical Update Support**: Immediate notifications for security-critical updates
  * **Flexible Control**: Environment variable and config file control (`ORBITRON_AUTO_UPDATE`)
* **Configuration Management Command**
  * **`orbitron config` Command**: Manage all application settings through CLI
  * **Subcommands**: `get`, `set`, `list` for comprehensive configuration control
  * **Auto-update Control**: Easy enable/disable of automatic updates
* **Documentation Quick Access**
  * **`/docs` Command**: Opens official documentation in browser
  * **Cross-platform Browser Support**: Works on Windows, macOS, Linux, and WSL
  * **Command Completion**: Integrated into autocomplete system

</details>

<details>

<summary>Improved</summary>

* **Persistent Bash Mode**
  * **Mode Persistence**: Bash mode (`!` prefix) now persists after command execution
  * **Explicit Exit**: Exit with ESC key or backspace on empty input
  * **Consecutive Commands**: No need to re-enter `!` for multiple commands
* **Hooks Configuration**
  * **Dedicated Hooks File**: Moved for better organization
  * **Automatic Migration**: Seamless migration from old config locations
  * **Global/Project Separation**: Clear distinction between global and project-specific hooks
* **Installation System**
  * **Cross-platform Scripts**: Bash and PowerShell installers
  * **Legacy Migration**: Automatic detection and migration from old installation paths
  * **Shell PATH Configuration**: Automatic shell detection and PATH setup on Unix, registry modification on Windows
  * **Test Infrastructure**: Separate test release system for safer update testing
* **Backend API Migration**
  * **Dedicated API Services**: Migrated from PostgREST to dedicated backend APIs
  * **Context API**: Context retrieval service with pagination support
  * **User Auth API**: Organization authentication service
  * **User Version API**: Client version update service
  * **Retry Logic**: Exponential backoff for improved reliability
  * **JWT Authentication**: Secure token-based authentication
* **File Handling**
  * **Escape Sequence Support**: Handles backslash-escaped paths from iTerm/macOS terminals
  * **Space in Filenames**: Fixed drag-and-drop issues with spaces in file names
* **Korean Input**
  * **Hangul Normalization**: NFC normalization prevents Jamo display corruption
  * **Optimized Detection**: Quick Jamo detection to skip normalization when not needed

</details>

<details>

<summary>Fixed</summary>

* **Configuration Data Preservation**
  * **Deep Merge Implementation**: Recursive merge for nested config maps (agents, providers, etc.)
  * **Agent Model Preservation**: All agent configurations preserved during model changes
  * **No Data Loss**: Fixed issue where updating individual fields would delete other entries
  * **Merge Strategy**: Non-nil, non-empty fields merged instead of overwritten
  * **Empty Field Cleanup**: Hooks field properly removed when last hook is deleted
* **Model Dialog Initialization**
  * **Unified Code Path**: Fixed "No Configured Providers" error when opening via `ctrl+o`
  * **Consistent Behavior**: Both `/model` command and keyboard shortcut now use same initialization
* **Resource Management**
  * **Response Body Cleanup**: Fixed defer-in-loop resource leak in retry logic
  * **Explicit Cleanup**: Per-branch response body cleanup in HTTP clients

</details>

## v0.1.12 - Interactive Bash Mode & Shell Hooks System - Oct 29, 2025

See what's new and improved in our October update.

<details>

<summary>New Features</summary>

* **Interactive Bash Mode with Cancellation Support**
  * **Direct Command Execution**: Enter/exit bash mode with `!` prefix for seamless command execution
  * **Asynchronous Operation**: Real-time command output display in chat with async execution
  * **Instant Cancellation**: Cancel running commands with ESC key
  * **Visual Indicators**: Clear status indicators for bash mode and command execution state
  * **Context Filtering**: System messages and bash history filtered from LLM context for cleaner conversations
  * **Thread-safe Management**: Robust state management across multiple concurrent sessions
* **Shell Hooks System for Tool Execution Control**
  * **Pre-execution Hooks**: `preToolUse` runs commands before tool execution with blocking capability (non-zero exit codes block execution)
  * **Post-execution Hooks**: `postMessageUse` runs commands after assistant completes response
  * **Flexible Targeting**: Target specific tools (e.g., bash, edit) or all tools with `*` wildcard
  * **Comprehensive Management**: `/hooks` command for full hook lifecycle (add/edit/delete/test)
  * **Rich UI**: Inline testing and real-time feedback in the interface
  * **Context Variables**: Extensive environment variables for hook scripts:
    * `ORBITRON_TOOL_NAME`: Name of the tool being executed
    * `ORBITRON_TOOL_INPUT`: JSON input to the tool
    * `ORBITRON_SESSION_ID`: Current session ID
    * `ORBITRON_MESSAGE_ID`: Current message ID
    * `ORBITRON_TOOL_RESULT`: Tool result (for postMessageUse hooks)
* **CLI Token-Based Authentication**
  * **Direct CLI Authentication**: Use `--access-token` and `--refresh-token` flags for direct authentication
  * **CI/CD Integration**: Streamlined authentication for automated workflows
  * **Token Validation**: Ensures both tokens are provided together for security

</details>

<details>

<summary>Improved</summary>

* **WSL OAuth Browser Opening**
  * Enhanced PowerShell method for more stable URL handling
  * Added `rundll32.exe` fallback when PowerShell fails
  * Improved WSL detection with `WSL_DISTRO_NAME` environment variable check
  * Better error handling and debug logging throughout OAuth flow
* **MCP Tools Stability**
  * Implemented buffered channels to prevent goroutine blocking
  * Added defer cleanup blocks for guaranteed session state release
  * Intelligent JSON summarization for large responses (150+ items)
  * Achieves up to 43% size reduction to prevent context overflow
  * Minimal MCP tool completion messages in UI for better readability

</details>

<details>

<summary>Fixed</summary>

* **WSL OAuth Flow**
  * Fixed issue where `&` character in OAuth URLs was interpreted as command separator
  * PowerShell as primary method prevents URL parsing issues
  * Enhanced error recovery with fallback mechanisms
* **WSL UNC Path File Attachments**
  * Full support for Windows UNC paths (`\\wsl.localhost\...` and `\\wsl$\...`)
  * Drag and drop files from Windows Explorer now works correctly for WSL file systems
  * Automatic path conversion from UNC to Linux paths
  * Proper handling of cross-distribution limitations
* **MCP Tools Session Management**
  * Fixed sessions getting stuck in "busy" state after MCP tool errors
  * Eliminated goroutine deadlocks during token refresh using TryLock pattern
  * Prevented context overflow errors (215,530 > 200,000 tokens) with intelligent summarization

</details>

***

## v0.1.11 - Dramatic TUI Performance & MCP Enhancements - Oct 17, 2025

See what's new and improved in our October update.

<details>

<summary>New Features</summary>

* **Dramatic TUI Performance Improvements**
  * **75x Faster Startup**: Reduced TUI initialization time from 3-5 seconds to 0.04 seconds by moving MCP server initialization to background
  * **Asynchronous MCP Initialization**: Separated TUI rendering from MCP loading to provide immediate user interface
  * **Context-based Lifecycle Management**: Prevents goroutine leaks and ensures clean shutdown through application context propagation
* **Model Registry System**
  * Implemented Model Registry system with dynamic model loading from API
  * Added support for global prefix in Bedrock cross-region inference
  * Improved retry logic to use true exponential backoff
* **Enhanced MCP Server Management**
  * Added visual status indicators for MCP servers (ready/loading/failed/disabled)
  * Implemented per-server status tracking functionality
  * Added MCP server enable/disable toggle with key binding support
  * Added disabled server status tracking for cache loads
  * Added configurable MCP initialization timeout constant
  * Added tool deduplication to prevent API errors
* **Token Tracking and Cost Management**
  * Added dedicated usage tracking table for token consumption
  * Added token and cost tracking to messages table
  * Added debug logging for tool counts in streaming requests
* **User Profile Enhancements**
  * Added client version tracking for user profiles
* **File Attachments**
  * Added WSL support for Windows path conversion in file attachments
  * Improved WSL path conversion robustness

</details>

<details>

<summary>Improved</summary>

* **MCP Initialization**
  * Initialize MCP tools before CloudEngineerAgent refresh
  * Enhanced npx resolution with dynamic cache detection and logging
  * Extracted default npx cache directory logic into helper function
  * Extracted common logic in MCP server state updates
  * Changed MCP status dot color from blue to green and extracted color constants
  * Improved MCP list dialog rendering and added status legend
* **Error Handling**
  * Handle invalid request errors from large attachments
  * Simplified invalid request error handling
  * Fixed tool\_use/tool\_result mismatch on mid-generation cancellation
  * Applied code review feedback for cancellation handling
* **Vertex AI**
  * Increased DefaultMaxTokens to 50,000 for Vertex AI Claude models
  * Separated max\_tokens detection from tool\_use in finish reason
* **Permission Dialog**
  * Improved type handling for WritePermissionsParams and EditPermissionsParams
* **Performance Optimization**
  * **75x Faster Startup**: 3-5 seconds → 0.04 seconds
  * Interface becomes immediately usable while MCP servers load asynchronously in background
  * Enhanced MCP server initialization performance with dynamic cache detection
  * Improved initialization stability with configurable timeouts
  * Improved API retry efficiency with true exponential backoff

</details>

<details>

<summary>Fixed</summary>

* **MCP Related**
  * Removed redundant MCP cache clear on toggle
  * Normalized environment variable keys to uppercase
  * Fixed API key loading after agent refresh
* **Resource Management**
  * Fixed goroutine leak: Application context propagation ensures proper cancellation of background tasks
  * Prevented memory leaks: MCP server connections and goroutines are properly cleaned up on app shutdown
* **File Attachments**
  * Improved WSL path conversion robustness
  * Removed original Windows path from text after file attachment in WSL
* **Database**
  * Used ErrorPersist for attachment removal DB update failure
* **Code Quality**
  * Cleaned up code and removed Korean comments
  * Addressed thread safety issues from code review
  * Addressed code review feedback

</details>

***

## v0.1.10 - Authentication & Environment Management Improvements - Sep 27, 2025

See what's new and improved in our September update.

<details>

<summary>New Features</summary>

* **Authentication & Environment Management**
  * Added environment detection and consistency validation for dev/prod environments
  * Implemented environment-aware OAuth URLs and authentication setup
  * Enhanced session environment tracking to prevent mismatched logins
  * Improved OAuth callback server with better error handling and port management
* **Configuration Improvements**
  * Refactored configuration system for better provider management from Supabase
  * Added dynamic provider loading with fresh initialization on reload
* **User Experience**
  * Replaced `/attachment` command with `Ctrl+G` keyboard shortcut for file attachment mode
  * Improved file path detection with debouncing for Windows drag-and-drop
  * Enhanced login dialog with better guidance and error messages
  * Fixed authentication state updates after successful login

</details>

<details>

<summary>Improved</summary>

* **Logging & Debugging**
  * Updated logging levels from Info to Debug for non-critical messages
  * Improved error logging for provider loading failures
  * Added session ID to metering data for better tracking

</details>

<details>

<summary>Fixed</summary>

* **Configuration & Authentication**
  * Fixed provider configuration reload issues after login
  * Resolved OAuth server port binding conflicts
* **File Handling**
  * Fixed Windows file path detection for drag-and-drop
  * Prevented false file attachment errors on quoted text paste
* **Code Quality**
  * Fixed merge conflicts and cleaned up debug logging

</details>

***

## v0.1.9 - Improving the TUI experience, session management - Sep 22, 2025

See what’s new and improved in our August update.

<details>

<summary>New Features</summary>

* **Model Context Protocol (MCP) Integration**

  * **MCP Server Management**: Added comprehensive MCP server support with dynamic tool updates and configuration management
  * **MCP Tools Caching**: Implemented caching mechanism to improve startup performance
  * **Status Monitoring**: Added `/mcp status` command for monitoring MCP server usage
  * **Enhanced Commands**: Improved MCP command parsing and help documentation

  **Session Management & Export**

  * **Session Permissions**: New session permissions management system with enhanced dialog functionality
  * **Export Functionality**: Added chat export feature with multiple format options (TXT, MD, JSON)
  * **Export Dialog**: New export dialog with format selection capabilities
  * **Batch Export**: Support for exporting entire chat history

  **File Handling Improvements**

  * **Drag & Drop**: Added drag-and-drop file attachment support
  * **Clipboard Paste**: Implemented clipboard paste functionality for file attachments
  * **Enhanced File Picker**: Improved file picker with CSV handling and encoding conversion

</details>

<details>

<summary>Improved</summary>

* **User Interface Enhancements**
  * **Permission Dialog**: Enhanced permission dialog with improved parameter handling
  * **Export Dialog**: New dedicated dialog for export operations
  * **Keyboard Shortcuts**: Added double ESC detection for text clearing
  * **Editor Improvements**: New ClearAll key binding for clearing text input

</details>

<details>

<summary>Fixed</summary>

* Auto Compact bug
* Token Counting bug

</details>

***

## v0.1.8 -Improving the TUI experience, session management - Sep 08, 2025

See what’s new and improved in our August update.

<details>

<summary>New Features</summary>

* **Code Signing & Distribution**
  * **Multi-platform Code Signing**: Full support for Windows, macOS, and Linux code signing
  * **macOS Notarization**: Integrated App Store Connect API for automatic notarization
  * **GPG Signing**: Added GPG signing support for Linux builds
  * **Windows Certificate Signing**: Implemented signtool-based signing for Windows executables
  * **Platform-specific Builds**: Separated GoReleaser configurations for optimized platform builds
* **Authentication & Security**
  * **Token Caching**: Implemented intelligent token caching for improved session management
  * **Auto-authentication**: Added support for automatic authentication mode
  * **Enhanced Error Handling**: Improved authentication retry logic and error recovery
  * **Session Persistence**: Better session management with cached credentials
* **Metering & Analytics**
  * **Token Usage Tracking**: Comprehensive metering service for all LLM providers
  * **Provider Support**: Added metering for Anthropic, OpenAI, Gemini, Bedrock, and VertexAI
  * **Detailed Logging**: Enhanced logging for usage analytics and debugging
  * **Cost Tracking**: Foundation for usage-based cost tracking

</details>

<details>

<summary>Improved</summary>

* **User Interface Improvements**
  * Enhanced File Picker:
    * Added image preview capabilities
    * Improved file type detection and handling
    * Better support for large directories
    * Size limit validations
  * Binary Content Support: Full support for binary content in chat messages
  * UTF-8 Safety: Improved string truncation with proper UTF-8 handling
  * Command Help: Refactored help dialog to display available slash commands
  * Version Update Notifications: Background version checking with update alerts
* **Performance**
  * Reduced authentication overhead through token caching
  * Optimized file picker performance for large directories
  * Improved message rendering with UTF-8 safety checks
  * Better memory management in binary content handling

</details>

<details>

<summary>Fixed</summary>

* Fixed title generation agent timeout issues with adjusted max tokens
* Resolved macOS certificate keychain installation issues
* Fixed binary content handling in various LLM providers
* Corrected UTF-8 string truncation edge cases

</details>

***

## v0.1.7 -Improving the TUI experience, session management - Aug 22, 2025

See what’s new and improved in our August update.

<details>

<summary>New Features</summary>

* Info Dialog:
  * **Session Information Display**: Introduced a new Info Dialog that provides users with easy access to session details including user email and current authentication mode
  * **Command Integration**: Added new `info` command to the TUI command parser with full completion support
  * **Enhanced User Experience**: Users can now quickly check their session status without leaving the current workflow
* Manual Compaction Command:
  * Added new `compact` command for users who want to manually trigger session compaction

</details>

<details>

<summary>Improved</summary>

* **Advanced Session Management**
  * **Intelligent Auto-Compaction: Implemented automatic session compacting that triggers at 85% of context window capacity, preventing context overflow issues**
  * **Smart Token Management: Added session information retrieval to check token count before summarization, ensuring safe operation within model limits**
  * **Message Truncation Logic: Introduced intelligent message truncation to manage token limits effectively while preserving conversation context**
  * **Streaming Summarization: Added streaming response handling for summarization process to prevent timeout issues and improve reliability**
* TUI Enhancements
  * **Enhanced Command System**: Extended command parser with new commands (`info`, `compact`) and updated command completion
  * **Improved Layout Management**: Enhanced chat page layout to handle Info Dialog display and closure seamlessly
  * **Better Help Documentation**: Updated help system to include new functionality and commands
  * **Streamlined User Interface**: Improved overall TUI responsiveness and user interaction flow

</details>

***

v0.1.6 -Improving the TUI experience, context management, and expanded AI\
model support including GPT-5 and Claude 4.1- Aug 12, 2025
----------------------------------------------------------

See what’s new and improved in our August update.

<details>

<summary>New Features</summary>

* Context Search Dialog: New interactive dialog for searching and managing contexts
* Enhanced Dialog System: Refactored dialog components with improved overlay rendering
* Slash Commands: Full slash command support in chat interface
* GPT-5 Family Support: Added support for latest GPT-5 models (gpt-5, gpt-5-nano, gpt-5-mini)
* Claude 4.1 Support: Added support for latest Claude models across multiple providers
* Improved Message Handling: Enhanced message retrieval and formatting in non-interactive mode

</details>

<details>

<summary>Improved</summary>

* Refined color schemes and terminal handling for better visibility
* Updated model configurations and dependencies
* Optimized context loading logic with better error handling

</details>

<details>

<summary>Fixed</summary>

* Fixed color visibility issues in Orbitron theme
* Resolved context loading edge cases
* Corrected dialog overlay rendering issues

</details>

***

## v0.1.5 -Improving the authentication experience, configuration management - Aug 06, 2025

See what’s new and improved in our August update.

<details>

<summary>Improved</summary>

* Enhance login dialog with refresh token support

</details>

***

## v0.1.4 -Improving the authentication experience, configuration management - Aug 06, 2025

See what’s new and improved in our July update.

<details>

<summary>New Features</summary>

**Authentication & User Management**

* **Enhanced Login System**: Added support for refresh tokens alongside access tokens for improved authentication
* **User Profile Display**: User email and authentication mode are now displayed in the session header
* **Improved Session Management**: Better integration with global database for enhanced user session handling

**AI Provider Support**

* **Google Vertex AI Integration**: Added comprehensive support for Google Vertex AI with service account authentication
* **Enhanced Model Configuration**: Improved handling of different AI models and providers
* **Better Error Handling**: More robust error messages and handling for API key configuration

**User Interface Improvements**

* **Enhanced Chat Editor**: Improved @ symbol handling and completion dialog functionality
* **Better Context Search**: Refined command parsing and context search dialog for improved functionality
* **Improved Text Rendering**: Added UTF-8 truncation support and better ANSI handling in diff rendering
* **Dynamic Editor Selection**: The editor command now dynamically selects available text editors

**API Key Management**

* **Enhanced API Keys Dialog**: Added pagination and required field validation for better user experience
* **Google Vertex AI Configuration**: Added specific configuration options for Google Vertex AI provider

</details>

<details>

<summary>Improved</summary>

* **Configuration Management**: Streamlined configuration handling and removed obsolete components

</details>

<details>

<summary>Fixed</summary>

* **Authentication Sync**: Fixed synchronization issues with Supabase authentication calls
* **Model Configuration**: Updated default model settings and improved logging in agent prompts
* **Release Process**: Fixed tag retrieval method in release scripts for more reliable releases

</details>

***

## v0.1.3 -Improving the authentication experience, configuration management - July 29, 2025

See what’s new and improved in our July update.

<details>

<summary>Improved</summary>

🔐 Enhanced Authentication & OAuth Flow

* **Improved OAuth Callback UI**: Replaced hardcoded HTML with template-based OAuth callback pages
* **Unified Result Page**: Added comprehensive `oauth_result.html` for both success and error states
* **Enhanced Supabase Integration**: Updated Supabase client with improved credential management and version checks
* **Better Token Handling**: Simplified token validation process for new API keys
* **Responsive Design**: Support for both light and dark themes with proper text visibility

**⚙️ Configuration Management Improvements**

* **Local Configuration Support**: Introduced `LocalConfig` struct for project-specific settings
* **Enhanced Configuration Loading**: Improved `Load` function to read and merge local configuration with global settings
* **Better Configuration Updates**: Implemented `saveLocalConfig` and `updateLocalConfig` functions
* **Improved Logging**: Enhanced logging throughout configuration loading and updating processes
* **API Key Dialog Enhancements**: Refactored API key handling for better clarity and error management

</details>

<details>

<summary>Fixed</summary>

* Fixed configuration update issues with in-memory config handling
* Improved error handling in Supabase client operations

</details>

***

## v0.1.2 - Enhanced User Experience - July 25, 2025

See what’s new and improved in our July update.

<details>

<summary>Improved</summary>

* **Enhanced Context Management**
  * **Context Search System**: Advanced search functionality for project contexts
  * **Command System**: New `/context` command with search capabilities
  * **Context Selection Dialog**: Improved dialog with pagination and filtering
  * **Agent Refresh**: Automatic agent refresh when context is selected
  * **Project Guidelines**: Real-time loading of updated project guidelines
* **Authentication Enhancements**
  * **JWT Management**: Maintained JWT expiry functionality with improved reliability
  * **User Session Tracking**: Better session management with user association
* **Improved User Interface**
  * **Inline Command Hints**: Colorful command hints in the editor interface
  * **Command Help System**: Replaced dialog with inline hints for better UX
  * **Context Indicators**: Visual indicators for organization contexts
  * **Search Functionality**: Enhanced search with URL encoding and field-specific queries
  * **Session Management**: Auto-creation of new sessions when context is selected

</details>

<details>

<summary>Fixed</summary>

* Fixed context search functionality with proper URL encoding
* Improved error handling in authentication services
* Enhanced session management reliability
* Fixed command help dialog display issues

</details>

***

## v0.1.1 - Configuration Management Update - July 23, 2025

See what’s new and improved in our July update.

<details>

<summary>Improved</summary>

* Optimized database queries for better performance
* Reduced configuration file I/O operations
* Improved startup time with streamlined initialization

</details>

<details>

<summary>Fixed</summary>

* Fixed configuration persistence issues across application restarts
* Improved error handling for database connection failures
* Enhanced session management reliability

</details>

***

## v0.1.1 - Initial Release - July 22, 2025

**Orbitron v0.1.0** marks the initial release of our powerful terminal-based AI assistant designed specifically for cloud engineers and developers. This release introduces a comprehensive AI-powered terminal interface that brings intelligent assistance directly to your command line.

<figure><img src="/files/f9IGsPlVHngdHKc9e2ey" alt=""><figcaption></figcaption></figure>

**Multi-Provider AI Support**

* **OpenAI Models**: GPT-4.1, GPT-4.5 Preview, GPT-4o, O1, O3, O4 Mini
* **Anthropic Claude**: Claude 4 Sonnet/Opus, Claude 3.5 Sonnet/Haiku, Claude 3.7 Sonnet
* **Google AI**: Gemini 2.5, Gemini 2.0 Flash series
* **Groq**: High-performance models including Llama 4 Maverick, QWEN QWQ-32b

**Rich Terminal User Interface (TUI)**

* **Interactive Chat Interface**: Seamless AI conversation experience
* **Session Management**: Multiple conversation sessions with easy switching
* **Real-time Streaming**: Live AI response streaming with typing indicators
* **Keyboard Shortcuts**: Comprehensive shortcut system for power users

**Developer Tools Integration**

* **File Operations**: View, edit, search, and manage project files
* **Shell Integration**: Execute commands directly within the interface
* **Web Fetching**: Retrieve and analyze web content

**Authentication & Security**

* **Multi-Provider API Key Management**: Secure storage and validation
* **Token-based Authentication**: OAuth flow with refresh token support

**Non-Interactive Mode**

* **Script Integration**: Command-line prompt execution for automation
* **JSON Output Support**: Structured output for programmatic use
* **Quiet Mode**: Silent execution for CI/CD pipelines