Formic
by rickywo
Agentic task manager that turns goals into shipped code with human review
How It Works
Orchestrates agentic workflows that turn high-level goals into shipped code by coordinating Claude Code and Copilot CLI. Uses structured planning, parallel execution, and configurable human review checkpoints to break tasks into subtasks and run them concurrently. Key features include task decomposition, execution orchestration, and review hooks for human-in-the-loop validation. This approach aligns with the Hierarchical Multi-Agent Pattern to structure responsibilities and scopes across agents, and with the Sub-Agent Delegation Pattern to enable effective task handoffs where appropriate.
Why It Matters
When to Use
Teams prototyping multi-agent developer workflows that need task decomposition, parallel execution, and human verification before committing code. Consider leveraging the Sub-Agent Delegation Pattern to distribute subtasks effectively, and the Agent-to-Agent Protocol (A2A) to coordinate across agents.
Applications
- Decomposing a product feature into parallel coding subtasks and coordinating multiple code-writing agents
- Adding human review gates to autonomous code changes to reduce regressions
- Experimenting with agent delegation patterns and observing failure modes during prototype development