orchestrating-parallel-worklisted

# Orchestrating Parallel Work ## Overview This skill provides patterns for coordinating parallel agent execution in Claude Code workflows. Apply these patterns when tasks can run simultaneously without interdependencies, achieving 3-6x speedup and 80-90% context savings. ## When to Use - Launching multiple research or exploration agents - Implementing features with independent subtasks - Running diagnostics that check multiple layers - Processing batch operations with dependency graphs - Any workflow where "wait for A, then start B" isn't required ## Key Concepts ### Background Agents Agents launched with `run_in_background: true` execute in isolated context. The main conversation continues while they work. ``` task = Task( description: "Describe what agent does", prompt: "Detailed instructions", subagent_type: "agent-type", run_in_background: true ) # Returns immediately with task.id ``` ### Collecting Results Use `TaskOutput` to wait for completion: ``` result = TaskOutput(task_id: task.id, block: true) # Wait for completion status = TaskOutput(task_id: task.id, block: false) # Check without waiting ``` ### Task Dependencies Use `TaskUpdate` to set up dependencies between tasks: ``` TaskUpdate({ taskId: childTask.id, addBlockedBy: [parentTask.id] }) ``` ## Decision Logic ### Should I Parallelize? Ask these questions: 1. **Are tasks independent?** → If no, use sequential 2. **Will each task take >30 seconds?** → If no, sequential might be fas