swarm▌

Spawn isolated agents for parallel task execution with runtime-native team selection and fresh context per wave. \n \n Auto-selects spawn backend based on runtime: Claude Native Teams in Claude sessions, Codex sub-agents in Codex sessions, or inline fallback if multi-agent unavailable \n Executes tasks in dependency-ordered waves, spawning only unblocked tasks in parallel while respecting file ownership conflicts via pre-spawn manifest validation \n Implements Ralph Wiggum pattern: fresh worker

Orchestrates parallel task execution across up to 12 concurrent Sparky subagents using a rolling pool scheduler. \n \n Parses markdown plan files, extracts task definitions, and launches subagents continuously as slots open without waiting for batch completion \n Maintains canonical file paths and naming constraints across parallel tasks to prevent filename drift and cross-task conflicts \n Validates each subagent result, updates the plan file with completion logs, and immediately schedules the

Generates dependency-aware implementation plans optimized for parallel multi-agent execution. \n \n Creates atomic, independently executable tasks with explicit dependency declarations to maximize parallelization across agents \n Requires codebase investigation, documentation retrieval for external libraries, and clarifying questions before planning to eliminate ambiguity \n Structures plans with task IDs, dependency arrays, descriptions, file locations, and validation criteria; includes visual