What problem does it solve?
ADAM orchestrates multiple subordinate “mule” agents in parallel while keeping their instructions dense and constrained, so complex multi-part engineering work can be executed efficiently and consistently.
Core Features & Use Cases
- Parallel multi-agent orchestration: Spawns multiple mules via Claude Code’s Task tool, each in an isolated workspace directory.
- Instruction density discipline: Converts driver intent into lean, imperative, structured prompts using the provided instruction-density guidance.
- Linear issue workflow integration (field updates only): Fetches issues, assigns to the correct user, and updates status via MCP tools while explicitly avoiding Linear comments.
- Workspace lifecycle & tracking: Maintains
/Users/eve/workspace/AGENTS/MULES.md as the single source of truth and archives completed or orphaned workspaces.
Quick Start
Ask an AI agent to run the ADAM driver to manage parallel work on multiple Linear issues (e.g., “Adam, coordinate mules to fix LINEAR-123 and LINEAR-456 in parallel, update their assignees/statuses, and report commit SHAs plus PR URLs when done”).