Dispatch

This is a background worker orchestrator that lets you spawn AI agents to handle tasks without blocking your main session. Say "dispatch opus to fix the bug" and it writes a checklist plan, spawns a worker via CLI (Cursor, Claude, or Codex), and tracks progress through IPC files. The routing is smart: it parses your prompt for model names or aliases, auto-discovers available models, and handles config updates when things break. Notable is the worktree isolation for Claude backend workers, so multiple agents can work in parallel without stomping on each other's files. The whole point is delegating work that would otherwise keep you waiting, like refactors, reviews, or research tasks. Config lives in ~/.dispatch/config.yaml and you can add models or aliases on the fly.