OpenCode Workspace

tip

450 GitHub stars — Bundled multi-agent orchestration harness. 16 components, one install, complete control.

Overview

OpenCode Workspace is a complete orchestration harness designed to make complex multi-agent workflows manageable. It bundles 16 components including agent delegation, worktree management, notifications, background agents, context management, and more into a single plug-and-play package.

What's Included

Component	Purpose
Agent Orchestrator	Coordinates task distribution across agents
Worktree Manager	Git worktree isolation (standalone or paired with opencode-worktree)
Notification System	Event-based desktop and sound notifications
Background Agent Runner	Async agent execution with result collection
Context Manager	Cross-session context persistance
Terminal Manager	Multi-terminal session tracking
Session Logger	Full session recording and replay
Permission Manager	Granular cross-agent permission controls
Health Monitor	Agent performance and error tracking
Task Scheduler	Timed and event-triggered task execution
Output Collector	Aggregated result collection from parallel agents
Conflict Detector	Identifies overlapping file modifications
State Manager	Shared state across agent sessions
Rate Limiter	Prevents API token exhaustion
Recovery Handler	Automatic retry and fallback on agent failure
Metrics Exporter	OpenTelemetry-compatible metrics

When to Use

Scenario	Benefit
Complex multi-agent workflows	Single config manages all agents and their interactions
Production deployments	Built-in safety rails and conflict detection
Team use	Consistent behavior across multiple developers
CI/CD integration	Reproducible agent pipelines

Installation

npm

{
  "plugin": ["opencode-workspace"]
}

Local

git clone https://github.com/kdcokenny/opencode-workspace.git ~/.config/opencode/plugins/opencode-workspace

Configuration

Base Configuration

{
  "plugin": ["opencode-workspace"],
  "opencode-workspace": {
    "orchestrator": {
      "max_concurrent_agents": 3,
      "timeout": 300000
    },
    "notifications": {
      "enabled": true,
      "sounds": false
    },
    "worktree": {
      "enabled": true,
      "cleanup_on_exit": true
    }
  }
}

Agent Configuration

{
  "agent": {
    "worker-1": {
      "description": "Primary implementation agent",
      "mode": "subagent",
      "permission": {
        "edit": "allow",
        "bash": "allow"
      }
    },
    "worker-2": {
      "description": "Secondary review agent",
      "mode": "subagent",
      "permission": {
        "edit": "deny",
        "read": "allow"
      }
    }
  }
}

Usage

Deploy a Multi-Agent Workflow

The orchestrator distributes tasks across configured agents:

/deploy-workflow implement and review the new authentication module

The orchestrator breaks this into subtasks, assigns them to available agents, monitors progress, and collects results.

Monitor Agent Health

/workspace status

Shows all active agents, their current tasks, resource usage, and status.

Best Practices

Start Small

• Begin with the core orchestrator and add components gradually.
• Each enabled component adds to the agent's context and tool surface.

Configure Timeouts

• Set reasonable timeouts for agent tasks to prevent runaway execution.
• Use the recovery handler to define fallback behavior on failure.

Cross-Agent Conflict Prevention

• Enable the Conflict Detector when multiple agents modify the same files.
• Use worktree isolation for agents that need full write access.

GitHub

• Repo: kdcokenny/opencode-workspace
• npm: opencode-workspace
• Topics: multi-agent, orchestration, harness, bundle
• Category: Agent Workflow / Orchestration