What problem does it solve?
Manually coordinating multiple parallel AI agent sessions across different harnesses like Claude Code, Codex, and OpenCode is tedious, error-prone, and wastes time switching between terminals and merging disjointed work.
Core Features & Use Cases
- Parallel Agent Management: Use dmux to create and manage multiple tmux panes, each running an independent AI agent session for supported harnesses including Claude Code, Codex, OpenCode, Cline, Gemini, and Qwen.
- Built-in Workflow Patterns: Leverage pre-defined templates for common multi-agent workflows including research + implementation, multi-file feature development, test-fix loops, cross-harness task assignment, and parallel code review.
- Use Case: For a full-stack feature build, assign one agent to handle database schema and migrations, a second to build API endpoints, and a third to create UI components, then merge all results into a single cohesive implementation.
Quick Start
Use the dmux-workflows skill to split a complex development task into parallel agent tracks for independent workstreams, then merge all outputs into a unified final result.