swarm-coordination

Multi-agent coordination in OpenCode breaks complex work into independent subtasks that run concurrently. You initialize a swarm planning flow, spawn parallel workers for multi-file changes, and monitor progress through persistent logs and post-completion reviews.

File reservation conflicts across parallel agents are managed through coordinated spawning workflows. The swarm coordination skill assigns file reservations to each worker, tracks progress with persistent logs, and performs reviews to minimize conflicts and maximize throughput.

Multi-agent swarm workflows are needed when shipping features with multiple components across a shared codebase, such as frontend, API, and tests. Tasks that benefit from decomposition, parallel execution, and cross-agent collaboration across files and milestones are ideal candidates.

OpenCode swarm coordination requires integration with Swarm Mail and hive-like task tracking. These components provide the persistent logging and cross-agent communication needed to manage subagent orchestration and track progress across milestones.

Workload balancing across spawned subagents is supported through worker orchestration. The coordination skill manages subagent spawning, assigns tasks based on decomposition, and balances workloads while tracking progress with persistent logs and file reservations.

Swarm coordination is not suited for single-file or linear tasks that lack decomposition opportunities. It is designed for parallel execution and cross-agent collaboration across multiple files and milestones, so simpler tasks gain minimal throughput benefit from spawning workers.