What problem does it solve? Coordinating multi-agent software work requires both upfront planning (RFCs, BRDs, ADRs, triage reports, board tasks) and isolated parallel execution; doing these by hand risks agents working from missing docs, conflicting branches, or unapproved plans. ## Core Features & Use Cases - Two-phase orchestration: Runs the full planning chain for new-feature, tech-task, investigate-bug, or investigate-crash in a single planning worktree, then blocks on explicit user approval before dispatching. - Parallel worktree dispatch: Creates one git worktree per implementation task from the resolved base branch and hands off to specialized agents in parallel with a mandatory setup preamble. - Artifact landing gate: Commits and merges Phase 1 docs and board tasks to the base branch, then verifies every doc and task exists on origin before any worktree is cut. - Use Case: Say you want to add a checkout feature spanning Android, iOS, and backend. Run /dispatch-task --type new-feature "checkout flow" to produce the RFC/BRD/ADR, create board tasks, and after approval dispatch three agents into isolated worktrees that each open PRs back to the base branch. ## Quick Start Ask the AI to run /dispatch-task --type new-feature with a short description of the feature you want planned and implemented in parallel.
