multi-agent-coordination

Multi-agent coordination uses git worktree isolation and file locking strategies to prevent conflicts when running multiple AI agents on the same codebase. Each agent operates in an isolated worktree with shared state management and inter-agent messaging to ensure safe parallel development.

Agent orchestration for parallel development involves task decomposition, message passing between agents, and a shared coordination document. This skill provides scaffolding for conflict detection, resolution, and patterns drawn from popular multi-agent developer tools like Claude Code, Cursor, and Devin.

Yes, git worktree isolation is the primary mechanism for enabling concurrent agents. You set up separate worktrees for each agent and initialize coordination using a shared coordination document, allowing agents to work in parallel with conflict-free file access and shared state management.

Inter-agent messaging enables coordination through lightweight message passing, conflict detection, and resolution scaffolding. Agents communicate via shared state management and a coordination document, supporting orchestrated parallel development with dead-agent salvage capabilities.

Multi-agent coordination is not suited for single-agent behavior patterns, agent infrastructure selection, or DAG topology design. If your task involves only one agent or requires infrastructure topology planning rather than parallel execution coordination, alternative approaches are more appropriate.

Multi-agent coordination includes file locking strategies and conflict resolution scaffolding to handle concurrent edits. The system detects conflicts between agents working on shared files and applies resolution patterns to maintain codebase integrity across isolated worktrees.