Implements regex-based response parser to extract file creation actions
and artifacts from LLM text responses. Agents can now produce actual
work products (files, PRs) instead of just returning instructions.
Changes:
- pkg/ai/response_parser.go: New parser with 4 extraction patterns
* Markdown code blocks with filename comments
* Inline backtick filenames followed by "content:" and code blocks
* File header notation (--- filename: ---)
* Shell heredoc syntax (cat > file << EOF)
- pkg/execution/engine.go: Skip sandbox when SandboxType empty/none
* Prevents Docker container errors during testing
* Preserves artifacts from AI response without sandbox execution
- pkg/ai/{ollama,resetdata}.go: Integrate response parser
* Both providers now parse LLM output for extractable artifacts
* Fallback to task_analysis action if no artifacts found
- internal/runtime/agent_support.go: Fix AI provider initialization
* Set DefaultProvider in RoleModelMapping (prevents "provider not found")
- prompts/defaults.md: Add Rule O for output format guidance
* Instructs LLMs to format responses for artifact extraction
* Provides examples and patterns for file creation/modification
* Explains pipeline: extraction → workspace → tests → PR → review
Test results:
- Before: 0 artifacts, 0 files generated
- After: 2 artifacts extracted successfully from LLM response
- hello.sh (60 bytes) with correct shell script content
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
Integrate chorus-dev-images repository with automatic language detection
and appropriate development container selection.
New features:
- ImageSelector for automatic language-to-image mapping
- Language detection from task context, description, and repository
- Standardized workspace environment variables
- Support for 7 development environments (Rust, Go, Python, Node, Java, C++)
Changes:
- pkg/execution/images.go (new): Image selection and language detection logic
- pkg/execution/engine.go: Modified createSandboxConfig to use ImageSelector
This ensures agents automatically get the right tools for their tasks without
manual configuration.
Related: https://gitea.chorus.services/tony/chorus-dev-images🤖 Generated with Claude Code (https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
This commit implements Phase 3 of the CHORUS task execution engine development plan,
replacing the mock implementation with a real AI-powered task execution system.
## Major Components Added:
### TaskExecutionEngine (pkg/execution/engine.go)
- Complete AI-powered task execution orchestration
- Bridges AI providers (Phase 1) with execution sandboxes (Phase 2)
- Configurable execution strategies and resource management
- Comprehensive task result processing and artifact handling
- Real-time metrics and monitoring integration
### Task Coordinator Integration (coordinator/task_coordinator.go)
- Replaced mock time.Sleep(10s) implementation with real AI execution
- Added initializeExecutionEngine() method for setup
- Integrated AI-powered execution with fallback to mock when needed
- Enhanced task result processing with execution metadata
- Improved task type detection and context building
### Key Features:
- **AI-Powered Execution**: Tasks are now processed by AI providers with appropriate role-based routing
- **Sandbox Integration**: Commands generated by AI are executed in secure Docker containers
- **Artifact Management**: Files and outputs generated during execution are properly captured
- **Performance Monitoring**: Detailed metrics tracking AI response time, sandbox execution time, and resource usage
- **Fallback Resilience**: Graceful fallback to mock execution when AI/sandbox systems are unavailable
- **Comprehensive Error Handling**: Proper error handling and logging throughout the execution pipeline
### Technical Implementation:
- Task execution requests are converted to AI prompts with contextual information
- AI responses are parsed to extract executable commands and file artifacts
- Commands are executed in isolated Docker containers with resource limits
- Results are aggregated with execution metrics and returned to the coordinator
- Full integration maintains backward compatibility while adding real execution capability
This completes the core execution engine and enables CHORUS agents to perform real AI-powered task execution
instead of simulated work, representing a major milestone in the autonomous agent capability.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
