Andy
f28d22984d
* fix(pr-review): properly capture structured output from SDK ResultMessage Fixes critical bug where PR follow-up reviews showed "0 previous findings addressed" despite AI correctly analyzing resolution status. Root Causes Fixed: 1. sdk_utils.py - ResultMessage handling - Added proper check for msg.type == "result" per Anthropic SDK docs - Handle msg.subtype == "success" for structured output capture - Handle error_max_structured_output_retries error case - Added visible logging when structured output is captured 2. parallel_followup_reviewer.py - Silent fallback prevention - Added warning logging when structured output is missing - Added _extract_partial_data() to recover data when Pydantic fails - Prevents complete data loss when schema validation has minor issues 3. parallel_followup_reviewer.py - CI status enforcement - Added code enforcement for failing CI (override to BLOCKED) - Added enforcement for pending CI (downgrade READY_TO_MERGE) - AI prompt compliance is no longer the only safeguard 4. test_dependency_validator.py - macOS compatibility fixes - Fixed symlink comparison issue (/var vs /private/var) - Fixed case-sensitivity comparison for filesystem Impact: Before: AI analysis showed "3/4 resolved" but summary showed "0 resolved" After: Structured output properly captured, fallback extraction if needed Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(pr-review): rescan related files after worktree creation Related files were always returning 0 because context gathering happened BEFORE the worktree was created. For fork PRs or PRs with new files, the files don't exist in the local checkout, so the related files lookup failed. This fix: - Adds `find_related_files_for_root()` static method to ContextGatherer that can search for related files using any project root path - Restructures ParallelOrchestratorReviewer.review() to create the worktree FIRST, then rescan for related files using the worktree path, then build the prompt with the updated context Now the PR review will correctly find related test files, config files, and type definitions that exist in the PR. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(pr-review): add visible logging for worktree creation and rescan Add always-visible logs (not gated by DEBUG_MODE) to show: - When worktree is created for PR review - Result of related files rescan in worktree This helps verify the fix is working and diagnose issues. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * feat(pr-review): show model name when invoking specialist agents Add model information to agent invocation logs so users can see which model each agent is using. This helps with debugging and monitoring. Example log output: [ParallelOrchestrator] Invoking agent: logic-reviewer [sonnet-4.5] [ParallelOrchestrator] Invoking agent: quality-reviewer [sonnet-4.5] Added _short_model_name() helper to convert full model names like "claude-sonnet-4-5-20250929" to short display names like "sonnet-4.5". Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(sdk-utils): add model info to AssistantMessage tool invocation logs The agent invocation log was missing the model info when tool calls came through AssistantMessage content blocks (vs standalone ToolUseBlock). Now both code paths show the model name consistently. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * feat(sdk-utils): add user-visible progress and activity logging Previously, most SDK stream activity was hidden behind DEBUG_MODE, making it hard for users to see what's happening during PR reviews. Changes: - Add periodic progress logs every 10 messages showing agent count - Show tool usage (Read, Grep, etc.) not just Task calls - Show tool completion results with brief preview - Model info now shown for all agent invocation paths Users will now see: - "[ParallelOrchestrator] Processing... (20 messages, 4 agents working)" - "[ParallelOrchestrator] Using tool: Read" - "[ParallelOrchestrator] Tool result [done]: ..." - "[ParallelOrchestrator] Invoking agent: logic-reviewer [opus-4.5]" Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(pr-review): improve worktree visibility and fix log categorization 1. Frontend log categorization: - Add "PRReview" and "ClientCache" to analysisSources - [PRReview] logs now appear in "AI Analysis" section instead of "Synthesis" 2. Enhanced worktree logging: - Show file count in worktree creation log - Display PR branch HEAD SHA for verification - Format: "[PRReview] Created temporary worktree: pr-xxx (1,234 files)" 3. Structured output detection: - Also check for msg_type == "ResultMessage" (SDK class name) - Add diagnostic logging in DEBUG mode to trace ResultMessage handling Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * chore(deps): update claude-agent-sdk to >=0.1.19 Update to latest SDK version for structured output improvements. Previous: >=0.1.16 Latest available: 0.1.19 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * refactor(sdk-utils): consolidate structured output capture to single location BREAKING: Simplified structured output handling to follow official Python SDK pattern. Before: 5 different capture locations causing "Multiple StructuredOutput blocks" warnings After: 1 capture location using hasattr(msg, 'structured_output') per official docs Changes: - Remove 4 redundant capture paths (ToolUseBlock, AssistantMessage content, legacy, ResultMessage) - Single capture point: if hasattr(msg, 'structured_output') and msg.structured_output - Skip duplicates silently (only capture first one) - Keep error handling for error_max_structured_output_retries - Skip logging StructuredOutput tool calls (handled separately) - Cleaner, more maintainable code following official SDK pattern Reference: https://platform.claude.com/docs/en/agent-sdk/structured-outputs Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * feat(pr-logs): enhance log visibility and organization for agent activities - Introduced a new logging structure to categorize agent logs into groups, improving readability and user experience. - Added functionality to toggle visibility of agent logs and orchestrator tool activities, allowing users to focus on relevant information. - Implemented helper functions to identify tool activity logs and group entries by agent, enhancing log organization. - Updated UI components to support the new log grouping and toggling features, ensuring a seamless user interface. This update aims to provide clearer insights into agent activities during PR reviews, making it easier for users to track progress and actions taken by agents. * fix(pr-review): address PR review findings for reliability and UX - Fix CI pending check asymmetry: check MERGE_WITH_CHANGES verdict - Add file count limit (10k) to prevent slow rglob on large repos - Extract CONFIG_FILE_NAMES constant to fix DRY violation - Fix misleading "agents working" count by tracking completed agents - Add i18n translations for agent activity logs (en/fr) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(pr-logs): categorize Followup logs to context phase for follow-up reviews The Followup source logs context gathering work (comparing commits, finding changed files, gathering feedback) not analysis. Move from analysisSources to contextSources so follow-up review logs appear in the correct phase. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Auto Claude
Autonomous multi-agent coding framework that plans, builds, and validates software for you.
Download
Stable Release
| Platform | Download |
|---|---|
| Windows | Auto-Claude-2.7.4-win32-x64.exe |
| macOS (Apple Silicon) | Auto-Claude-2.7.4-darwin-arm64.dmg |
| macOS (Intel) | Auto-Claude-2.7.4-darwin-x64.dmg |
| Linux | Auto-Claude-2.7.4-linux-x86_64.AppImage |
| Linux (Debian) | Auto-Claude-2.7.4-linux-amd64.deb |
| Linux (Flatpak) | Auto-Claude-2.7.4-linux-x86_64.flatpak |
Beta Release
⚠️ Beta releases may contain bugs and breaking changes. View all releases
| Platform | Download |
|---|---|
| Windows | Auto-Claude-2.7.2-beta.10-win32-x64.exe |
| macOS (Apple Silicon) | Auto-Claude-2.7.2-beta.10-darwin-arm64.dmg |
| macOS (Intel) | Auto-Claude-2.7.2-beta.10-darwin-x64.dmg |
| Linux | Auto-Claude-2.7.2-beta.10-linux-x86_64.AppImage |
| Linux (Debian) | Auto-Claude-2.7.2-beta.10-linux-amd64.deb |
| Linux (Flatpak) | Auto-Claude-2.7.2-beta.10-linux-x86_64.flatpak |
All releases include SHA256 checksums and VirusTotal scan results for security verification.
Requirements
- Claude Pro/Max subscription - Get one here
- Claude Code CLI -
npm install -g @anthropic-ai/claude-code - Git repository - Your project must be initialized as a git repo
Quick Start
- Download and install the app for your platform
- Open your project - Select a git repository folder
- Connect Claude - The app will guide you through OAuth setup
- Create a task - Describe what you want to build
- Watch it work - Agents plan, code, and validate autonomously
Features
| Feature | Description |
|---|---|
| Autonomous Tasks | Describe your goal; agents handle planning, implementation, and validation |
| Parallel Execution | Run multiple builds simultaneously with up to 12 agent terminals |
| Isolated Workspaces | All changes happen in git worktrees - your main branch stays safe |
| Self-Validating QA | Built-in quality assurance loop catches issues before you review |
| AI-Powered Merge | Automatic conflict resolution when integrating back to main |
| Memory Layer | Agents retain insights across sessions for smarter builds |
| GitHub/GitLab Integration | Import issues, investigate with AI, create merge requests |
| Linear Integration | Sync tasks with Linear for team progress tracking |
| Cross-Platform | Native desktop apps for Windows, macOS, and Linux |
| Auto-Updates | App updates automatically when new versions are released |
Interface
Kanban Board
Visual task management from planning through completion. Create tasks and monitor agent progress in real-time.
Agent Terminals
AI-powered terminals with one-click task context injection. Spawn multiple agents for parallel work.
Roadmap
AI-assisted feature planning with competitor analysis and audience targeting.
Additional Features
- Insights - Chat interface for exploring your codebase
- Ideation - Discover improvements, performance issues, and vulnerabilities
- Changelog - Generate release notes from completed tasks
Project Structure
Auto-Claude/
├── apps/
│ ├── backend/ # Python agents, specs, QA pipeline
│ └── frontend/ # Electron desktop application
├── guides/ # Additional documentation
├── tests/ # Test suite
└── scripts/ # Build utilities
CLI Usage
For headless operation, CI/CD integration, or terminal-only workflows:
cd apps/backend
# Create a spec interactively
python spec_runner.py --interactive
# Run autonomous build
python run.py --spec 001
# Review and merge
python run.py --spec 001 --review
python run.py --spec 001 --merge
See guides/CLI-USAGE.md for complete CLI documentation.
Development
Want to build from source or contribute? See CONTRIBUTING.md for complete development setup instructions.
For Linux-specific builds (Flatpak, AppImage), see guides/linux.md.
Security
Auto Claude uses a three-layer security model:
- OS Sandbox - Bash commands run in isolation
- Filesystem Restrictions - Operations limited to project directory
- Dynamic Command Allowlist - Only approved commands based on detected project stack
All releases are:
- Scanned with VirusTotal before publishing
- Include SHA256 checksums for verification
- Code-signed where applicable (macOS)
Available Scripts
| Command | Description |
|---|---|
npm run install:all |
Install backend and frontend dependencies |
npm start |
Build and run the desktop app |
npm run dev |
Run in development mode with hot reload |
npm run package |
Package for current platform |
npm run package:mac |
Package for macOS |
npm run package:win |
Package for Windows |
npm run package:linux |
Package for Linux |
npm run package:flatpak |
Package as Flatpak (see guides/linux.md) |
npm run lint |
Run linter |
npm test |
Run frontend tests |
npm run test:backend |
Run backend tests |
Contributing
We welcome contributions! Please read CONTRIBUTING.md for:
- Development setup instructions
- Code style guidelines
- Testing requirements
- Pull request process
Community
- Discord - Join our community
- Issues - Report bugs or request features
- Discussions - Ask questions
License
AGPL-3.0 - GNU Affero General Public License v3.0
Auto Claude is free to use. If you modify and distribute it, or run it as a service, your code must also be open source under AGPL-3.0.
Commercial licensing available for closed-source use cases.
