What problem does it solve?
When a task naturally splits into multiple independent domains, the main conversation context can become noisy, slow, and unwieldy; this Skill provides a protocol to dispatch isolated subagents so independent work runs concurrently and returns narrowly structured results ready for integration.
Core Features & Use Cases
- Scoped parallelization: Run 2+ isolated Claude instances, each with a single narrowly defined goal to avoid context bloat.
- Briefing checklist: Enforce a required brief structure (one-sentence goal, necessary context, constraints, explicit output contract, and optional length cap) so subagents produce verifiable outputs.
- Dispatch rules & safety: Batch tool calls to ensure concurrency, partition work by file/module to avoid edit conflicts, and prevent subagent recursion.
- Use Cases: Finding call sites across modules (one agent per symbol), researching multiple libraries concurrently (one agent per library), or auditing tests across packages (one agent per package).
Quick Start
Brief three read-only subagents with one-sentence goals to find all file:line matches for the symbols X, Y, and Z and return a concise bullet list of matches.
