Releases: unit-mesh/auto-dev
Compose Release compose-0.1.6
AutoDev Compose Multiplatform Release Test
This is a test release for the mpp modules containing:
Server
- mpp-server fat JAR (ready to run with
java -jar mpp-server-*-all.jar)
Android
- Debug APK
- Release APK
Desktop
- Linux (.deb package)
- Windows (.msi installer)
- macOS (.dmg image)
Note: This is a test release. Please use with caution.
Compose Release compose-0.1.5
🚀 AutoDev Compose 0.1.5 Release Notes
This release focuses on stability, cross-platform agent execution, and an improved Compose-based agent UI. Significant enhancements were made across agent functionality, multi-platform shell execution, chat workflows, and Android support. Numerous bug fixes improve reliability, tokenizer handling, Gradle/CI setup, and UI rendering.
✨ New Features
Agent & Automation
- Introduced the CodingAgent architecture for autonomous coding workflows.
- Added sub-agent management, including ErrorRecoveryAgent, LogSummaryAgent, and CodebaseInvestigatorAgent.
- Added AgentChannel, AgentEvent, submission model, and conversation management.
- Implemented tool discovery, filtering, and structured categorization.
- Added recovery advice rendering, improved action flow, and better LLM response parsing.
- Implemented customizable output rendering with improved formatting.
User Interface (Compose UI)
- Added a dedicated agent chat interface (Compose-based).
- Added copy-all / block copy, full output toggles, and parameter visibility switches.
- Introduced project explorer sidebar with resizable split panes.
- Added structured chat message display with streaming support.
Chat & Prompt Handling
- Chat history persistence and improved input routing.
- Support for @ mentions, variable prompts, and command-based input routing.
Platform Support
- Implemented PTY Shell Executor.
- Introduced Android SAF-based file system, file chooser, and Android shell execution support.
- Added Node.js target for CLI builds.
🐞 Bug Fixes
- Improved lexer parsing for markdown, variables, and special characters.
- Resolved issues with timestamp handling, max token length, and home directory fallback.
- Added stability fixes for shell process output handling.
- Fixed issues with Gradle Java home resolution and Kotlin compiler compatibility.
- Improved CI workflows, including Android SDK setup and Compose runtime test workflow.
🧱 Build & CI Improvements
- CI now uses pre-installed Android SDK on GitHub Actions.
- Added Compose release test workflow.
- Gradle config updated for consistent Java 21 toolchain handling.
- Automatic version increment support for
compose-*release tagging.
V2.4.6 - Spec Driven Development
AutoDev now supports GitHub Spec-Kit, a powerful toolkit for Spec-Driven Development that helps you build high-quality software faster by focusing on product scenarios and predictable outcomes.
What is Spec-Driven Development?
Spec-Driven Development flips the traditional approach - specifications become executable, directly generating working implementations rather than just guiding them. Instead of "vibe coding" every piece from scratch, you define the "what" and "why" first, then let AI agents handle the "how".
Key Features:
- Automatic Command Discovery: AutoDev automatically discovers and loads SpecKit commands from
.github/prompts/directory - Dynamic Command Execution: Use
/speckit.<subcommand>commands directly in DevIns scripts - Template Variable Support: Advanced template compilation with YAML frontmatter and variable resolution
- Code Completion: Full IntelliSense support for all SpecKit commands
Available Commands:
/speckit.constitution- Create or update project governing principles and development guidelines/speckit.specify- Define what you want to build (requirements and user stories)/speckit.plan- Create technical implementation plans with your chosen tech stack/speckit.tasks- Generate actionable task lists for implementation/speckit.implement- Execute all tasks to build the feature according to the plan/speckit.clarify- Clarify underspecified areas (recommended before planning)/speckit.analyze- Cross-artifact consistency & coverage analysis/speckit.checklist- Generate custom quality checklists
How It Works:
- SpecKit commands are loaded from
.github/prompts/directory in your project - Each command file follows the pattern
speckit.<subcommand>.prompt.md - Commands support YAML frontmatter for variable definitions
- Templates are compiled using Velocity engine with full variable resolution
Example Usage:
/speckit.constitution Create principles focused on code quality, testing standards, and performance
/speckit.specify Build a photo album application that organizes photos by date with drag-and-drop support
/speckit.plan Use React with TypeScript, Vite, and local SQLite database
2.4.6 (2025-10-28)
Bug Fixes
- DevInsProcessProcessor: update stream output to use empty string for better compatibility (910ac29)
- usedprocessor: wrap command execution in runReadAction for thread safety (76e8776)
Features
- builtincommand: check for custom commands before warning on not found (b6c889b)
- ClaudeSkillCommand: enhance template compilation by adding input parameter (df10651)
- template: add fallback for Velocity engine in test mode (ff0cafe)
2.4.5 (2025-10-28)
Bug Fixes
- AgentsInsCommand: streamline error messages and improve request validation (245b34e)
Features
- A2AInsCommand: update command examples for clarity and add AgentRequest data class (3c0fd0e)
- AgentsInsCommand: enhance error handling for agent invocation and improve not found message (467a3c6)
- BaseLangDictProvider: optimize file and class weight calculations with separate ReadAction handling (a646674)
- claudeskill: add Claude Skill command support #452 (8e9742e)
- ContextManagement: implement context window management and token budget allocation for VCS changes (45e1eaa)
- ContextWindowManager: add custom factory method for token budget initialization (429b446)
- dataprovider: add ClaudeSkillCommand to all() results (97b61b5), closes #452
- DiffSimplifier: add check for excessive lines in file revisions (a32f44c)
- DomainDictGenerateAction: enhance CSV output cleaning and README inclusion logic based on token budget (134aa3d)
- indexer: improve domain dictionary extraction rules (21e1dd0)
- JavaLangDictProvider: extract Level 1 file semantic names (48e726d)
- JavaLangDictProvider: implement collectLevel2 method for class and method name extraction (f784279)
- KotlinLangDictProvider: add semantic name extraction and tests (2e00b2a)
- LangDictProvider: add maxTokenLength parameter to collectFileNames method (250f602)
- LangDictProvider: implement two-level semantic name collection for improved LLM context (4ec3431)
- speckit: add frontmatter parsing and template compiler #452 (1b9ec3c)
- speckit: add load SpecKit command support and completion #452 (56542e6)
- VcsPrompting: add hasChanges method to check for modifications (830e654)
- weighting: add file and class weight calculation for LLM (2483d3c)
v2.4.3
Features
- A2AInsCommand: implement A2A command for sending messages to agents #451 (2d088d0)
- add RunInsCommand for executing auto commands in a project (27acd17)
- AgentsInsCommand: add AgentRequest data class for command requests and update JSON parsing #451 (e9d345d)
- AgentsInsCommand: add command for listing and invoking AI agents and closed #451 (873c8e7)
- AgentsInsCommand: enhance usage examples and improve agent listing format #451 (79a3612)
- AgentsInsCommand: improve A2A agent message handling and error reporting #451 (2ca69ac)
- AgentsInsCommand: refactor usage examples and agent info formatting #451 (706d80f)
- CommandResolver: add command resolution functionality for Windows and Unix systems (cb89135)
- DiffPreview: add standalone editor tab diff preview functionality (708cd73)
- DiffViewerService: implement diff viewer service with standalone diff processing and preview capabilities (3928fc2)
- DiffViewerService: implement standalone diff preview functionality (c153e9a)
- ProjectAgentsMD: implement AGENTS.md support for project context and closed #450 (1e7642d)