What problem does it solve? Delegating work to multiple AI subagents often leads to overlapping edits, unclear ownership, and unverified results. This Skill defines a disciplined orchestration workflow so Codex subagents receive bounded tasks, explicit file ownership, and their combined output is actually inspected and tested before acceptance. ## Core Features & Use Cases - Bounded Delegation: Assign each agent one objective, relevant source pointers, explicit file/responsibility ownership, success checks, and a stop condition. - Safe Concurrency: Use isolated worktrees when concurrent edits overlap, and ensure editing agents preserve other contributors' work. - Verified Integration: The primary agent owns integration, reconciles changes, inspects the combined diff, and runs checks against the integrated result rather than trusting completion messages. - Use Case: When a user asks to parallelize a refactor across three modules, split the work into independent briefs, run agents in isolated worktrees, then merge and run the full test suite on the combined diff. ## Quick Start Ask the agent to split this task among Codex subagents with clear ownership and verify the integrated result before committing.
