Attune AI v4.4.0: Native Agent Orchestration for Claude Code
Attune AI v4.4.0 is here with native agent orchestration designed specifically for Claude Code users.
What's New in v4.4.0
Native Composition Patterns
We've replaced external dependencies with 6 built-in composition patterns:
- Sequential - Tasks executed in order (A then B then C)
- Parallel - Independent tasks run simultaneously
- Debate - Multiple perspectives synthesized into consensus
- Teaching - Cost-effective generation with expert validation
- Refinement - Iterative improvement through draft/review cycles
- Adaptive - Smart routing based on task complexity
No CrewAI. No LangGraph. Just native Python patterns that work out of the box.
10 Smart Wizards
Production-ready wizards for software development:
| Wizard | Domain | Description |
|---|---|---|
security-audit | Security | Vulnerability scanning, injection risks, compliance |
code-review | Quality | Best practices, maintainability, bug detection |
bug-predict | Debugging | Pattern-based bug prediction |
perf-audit | Performance | Bottleneck identification, optimization |
refactor-plan | Architecture | Safe refactoring strategies |
test-gen | Testing | Test case generation, coverage improvement |
doc-gen | Documentation | API docs, READMEs, guides |
dependency-check | Dependencies | Vulnerability and license auditing |
release-prep | Release | Pre-release quality gates |
research | Research | Information synthesis |
14 Integrated Workflows
10 base workflows plus 4 new meta-workflows:
- release-prep - Automated release preparation with health checks
- test-coverage-boost - Intelligent test generation to hit coverage targets
- test-maintenance - Keep tests healthy and relevant
- manage-docs - Documentation quality management
Claude Code Integration
Agent and team creation features require Claude Code. This includes:
- Socratic agent builder (guided agent creation)
- Dynamic agent composition
- Custom team orchestration
- Real-time agent spawning
Core workflows work with any provider (OpenAI, Gemini, Ollama), but the agent creation magic requires Claude.
Getting Started
# Install
pip install attune-ai
# Run a wizard
attune workflow run security-audit
# Run with context
attune workflow run test-gen --input '{"path": "./src"}'
# Check project health
attune workflow run release-prep
Why Native Patterns?
- Zero Dependencies - No pip install of external frameworks
- Faster Startup - No heavy framework initialization
- Better Debugging - Clear Python stack traces
- Full Control - Customize patterns without fighting abstractions
Upgrade from v4.3
pip install --upgrade attune-ai
If you were using CrewAI integration, see our migration guide for the native pattern equivalents.
Questions? Join the discussion on GitHub.
Related Articles
The Grammar of AI Collaboration: Building Dynamic Agent Teams
What if AI agents composed themselves like words form sentences? Introducing a composable system for multi-agent orchestration with 10 composition patterns.
Dynamic Agent Creation: Teaching AI to Spawn Itself
Inside the Agent Factory—where templates become task-specific specialists. How we built a system that spawns customized AI agents on demand.
Building Agent Teams: From Words to Sentences
How composition patterns transform individual agents into collaborative teams. The 6 core patterns for multi-agent orchestration.