Toolkit
Multi-Agent Coordination · Compositions
On this page6 sections
Compositions
Teams + Task Ledger
Use stable task IDs, ownership, status, dependencies, and acceptance criteria. Product-native lists are useful coordination aids; external workflow state is stronger when work must survive sessions or products.
Teams + Worktrees
Give writing workers isolated branches and a named integrator. A worktree reduces file collisions but does not resolve semantic conflicts.
Teams + Role Skills
Define the portable role contract once and attach specialized skills for review, testing, research, or migration. Keep product-specific AgentA worker you delegate to: brief it, and it takes steps on its own. A chatbot answers; an agent acts.An LLM that runs tools in a loop toward a goal — it acts, checks the result, decides the next step, and repeats until done.Full definition definitions thin.
Teams + Hooks
Record task start, completion, failure, and scope checks where the product exposes those events. Verify event coverage for child and peer actions.
Teams + Memory
Use shared, reviewed state for decisions every worker needs. Keep private role memory for calibration only; otherwise workers develop incompatible realities.
Teams + External Orchestrator
When retries, schedules, approvals, or auditability matter, let deterministic workflow state own the run. Agents propose and execute bounded steps; the orchestrator controls transition and recovery.
Source register
Multi-Agent Coordination sources
Primary vendor documentation reviewed on the date shown. Links may change after publication.
- Orchestrate teams of Claude Code sessions (opens in a new tab)
Anthropic · official docs · accessed 2026-08-01
- Subagents (opens in a new tab)
OpenAI · official docs · accessed 2026-08-01
- GitHub Copilot CLI command reference (opens in a new tab)
GitHub · official docs · accessed 2026-08-01
- Running tasks in parallel with Copilot CLI (opens in a new tab)
GitHub · official docs · accessed 2026-08-01