Run a solo multi-agent coding cockpit with dev-3.0
Use dev-3.0 to turn coding-agent work into isolated Kanban tasks with git worktrees, tmux sessions, live previews, multi-agent launches, and review handoff.
npx skills add agentskillexchange/skills --skill run-a-solo-multi-agent-coding-cockpit-with-dev-3-0
dev-3.0 is a Kanban-first cockpit for a solo operator running multiple AI coding agents across repositories and worktrees. An agent/operator creates a task, lets dev-3.0 provision an isolated git worktree, launches Claude Code, Codex, Gemini CLI, opencode, Cursor, or another shell agent in tmux, monitors live terminal previews, and brings the result through review before landing work.
What this skill actually does
Use this when the user is already operating several coding agents and needs a repeatable task-to-worktree-to-agent-to-review loop instead of manually juggling terminals, branches, and local dashboards. The boundary is solo coding-agent fleet operations around task isolation, tmux launch, live monitoring, and review handoff; it is not an IDE, general project-management platform, or hosted team orchestration product.
Inputs and prerequisites: dev-3.0, git, tmux, and at least one local coding agent such as Claude Code, Codex CLI, Gemini CLI, opencode, or Cursor.
Setup notes: For macOS, run brew tap h0x91b/dev3 && brew trust h0x91b/dev3 && brew install –cask dev3. For Linux remote mode, run brew tap h0x91b/dev3 && brew trust h0x91b/dev3 && brew install h0x91b/dev3/dev3, then start the browser UI with dev3 remote. Agents can also follow https://dev3.h0x91b.com/ai-install.txt.
Source and verification boundary: use https://dev3.h0x91b.com/ as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.