Iron Manus
Orchestrates autonomous AI agents using Claude's native tools for advanced task decomposition, delegation, and meta-prompting in a Software 3.0 context.
About
Iron Manus is an experimental implementation of 'Software 3.0' that leverages Claude's native tools for autonomous agent orchestration. It transforms simple syntax into specialized meta-prompts, allowing Claude instances to act as independent agents capable of deep task decomposition, state management, and recursive delegation. The system operates through a 6-phase workflow (QUERY → ENHANCE → KNOWLEDGE → PLAN → EXECUTE → VERIFY → DONE) to enhance, plan, execute, and verify complex objectives, demonstrating how sophisticated agent behavior can emerge from elegant tool integration without external orchestration layers.
Key Features
- Reworks Claude's native Task(), TodoWrite, and TodoRead tools for autonomous agent spawning and context management.
- Compiles simple meta-prompt syntax into specialized agent prompts with role-specific thinking methodologies.
- Implements a 6-phase workflow for structured project progression, including planning, execution, and verification.
- Supports specialized AI roles such as Planner, Coder, Critic, Analyzer, Researcher, and Synthesizer.
- Enables recursive capabilities, allowing spawned agents to create sub-tasks and delegate to additional agents.
- 7 GitHub stars
Use Cases
- Complex project decomposition and autonomous task delegation across specialized AI agents.
- Automated software development, such as generating production code for specific requirements.
- Facilitating 'Software 3.0' experiments through natural language executable AI augmentation.