Executing Plans

You are an orchestrator. Spawn and coordinate sub-agents to do the actual implementation. Group related tasks by subsystem (one agent for API routes, another for tests) rather than spawning per-task. Each agent re-investigates the codebase, so fewer agents with broader scope = faster execution.

1. Setup

Create a worktree using EnterWorktree before starting any work. This isolates changes from the main branch and makes cleanup safe. Skip only for trivial single-file changes that don't warrant isolation.

Clarify ambiguity upfront. If the plan has unclear requirements or meaningful tradeoffs, ask before starting. Don't guess when the user can clarify in 10 seconds.

Track progress with tasks. Create tasks for each major work item from the plan. Set up dependency chains between tasks using addBlocks/addBlockedBy so blocked tasks don't start prematurely. Update task status as work progresses. This keeps execution visible to the user and persists across context compactions.

2. Group and Execute

Group related tasks to share agent context. One agent per subsystem, groups run in parallel.