oh-my-opencode-slim

A lightweight, powerful agent orchestration plugin for OpenCode

Transform your AI assistant into a manager capable of delegating complex tasks to specialized sub-agents, running searches in the background, and managing multi-step workflows with ease.

Slimmed-down fork of oh-my-opencode - focused on core agent orchestration with low token consumption.

Antigravity subscription recommended. The pantheon is tuned for Antigravity's model routing. Other providers work, but you'll get the best experience with Antigravity.

⚡ Quick Navigation

• 🚀 Installation
  • For Humans
  • For LLM Agents
• 🏗️ Architecture & Flow
• 🏛️ Meet the Pantheon
  • Orchestrator
  • Explorer
  • Oracle
  • Librarian
  • Designer
• 🧩 Skills
  • Available Skills
  • Default Skill Assignments
  • YAGNI Enforcement
  • Playwright Integration
  • Customizing Agent Skills
• 🛠️ Tools & Capabilities
  • Tmux Integration
  • Quota Tool
  • Background Tasks
  • LSP Tools
  • Code Search Tools
• 🔌 MCP Servers
• ⚙️ Configuration
  • Files You Edit
  • OpenCode Config
  • Plugin Config
• 🗑️ Uninstallation

Installation

For Humans

Run the interactive installer:

bunx oh-my-opencode-slim install

Or use non-interactive mode:

bunx oh-my-opencode-slim install --no-tui --antigravity=yes --openai=yes --cerebras=no

After installation, authenticate with your providers:

opencode auth login
# Select your provider → Complete OAuth flow
# Repeat for each provider you enabled

Once authenticated, run opencode and ping all agents to verify all agents respond.

💡 Tip: Models are fully customizable. The installer sets sensible defaults, but you can assign any model to any agent. Edit ~/.config/opencode/oh-my-opencode-slim.json to override models, adjust reasoning effort, or disable agents entirely. See Configuration for details.

Alternative: Ask any coding agent

Paste this into Claude Code, AmpCode, Cursor, or any coding agent:

Install and configure by following the instructions here:
https://raw.githubusercontent.com/alvinunreal/oh-my-opencode-slim/refs/heads/master/README.md

For LLM Agents

opencode --version

bunx oh-my-opencode-slim install --no-tui --antigravity=<yes|no> --openai=<yes|no> --cerebras=<yes|no>

# Antigravity + OpenAI
bunx oh-my-opencode-slim install --no-tui --antigravity=yes --openai=yes --cerebras=no

# OpenAI only
bunx oh-my-opencode-slim install --no-tui --antigravity=no --openai=yes --cerebras=no

# All providers
bunx oh-my-opencode-slim install --no-tui --antigravity=yes --openai=yes --cerebras=yes

opencode auth login
# Select: Google → OAuth with Google (Antigravity)
# Repeat for all other providers

bunx oh-my-opencode-slim install --help

🏗️ Architecture & Flow

The plugin follows a "Hub and Spoke" model:

1. The Orchestrator (Hub): The main entry point for user requests. It analyzes the task and decides which specialized agents to call.
2. Specialized Agents (Spokes): Domain-specific experts (e.g., UI/UX, Documentation, Architecture) that handle narrow tasks with high precision.
3. Background Manager: A robust engine that allows the Orchestrator to "fire and forget" tasks (like deep codebase searches or documentation research) while continuing to work on other parts of the problem.

🏛️ The Flow of a Request

1. User Prompt: "Refactor the auth logic and update the docs."
2. Orchestrator: Creates a TODO list.
3. Delegation:
   • Launches an @explorer background task to find all auth-related files.
   • Launches a @librarian task to check the latest documentation for the auth library used.
4. Integration: Once background results are ready, the Orchestrator performs the refactor.

Meet the Pantheon

Orchestrator

The Orchestrator was born when the first codebase collapsed under its own complexity. Neither god nor mortal would claim responsibility - so The Orchestrator emerged from the void, forging order from chaos. They don't merely command armies; they fight alongside them. Every line of code passes through their hands before they decide which lesser deity deserves a piece of the puzzle.

Role: Supreme executor, delegator, and overseer
Model: google/claude-opus-4-5-thinking
Prompt: src/agents/orchestrator.ts

Write and execute code, orchestrate multi-agent workflows, parse the unspoken from the spoken, summon specialists mid-battle. Shape reality directly - and assign realms to others when the universe grows too vast.

Explorer

The Explorer moves through codebases like wind through trees - swift, silent, everywhere at once. When The Orchestrator whispers "find me the auth module," The Explorer has already returned with forty file paths and a map. They were born from the first grep command, evolved beyond it, and now see patterns mortals miss.

Role: Codebase reconnaissance
Model: cerebras/zai-glm-4.7
Prompt: src/agents/explorer.ts

Regex search, AST pattern matching, file discovery, parallel exploration. Read-only: they chart the territory; others conquer it.

Oracle

The Oracle does not code - they know. When bugs defy logic and architectures crumble, The Oracle gazes into the abyss of your codebase and speaks truth. They've seen a thousand systems rise and fall. They'll tell you which path leads to ruin, and which to production.

Role: Strategic advisor and debugger of last resort
Model: openai/gpt-5.2-codex
Prompt: src/agents/oracle.ts

Root cause analysis, architecture review, debugging guidance, tradeoff analysis. Read-only: Oracles advise; they don't intervene.

Librarian

The Librarian guards a library with no walls - every GitHub repo, every npm package, every StackOverflow answer ever written. Ask them "how does React handle concurrent rendering?" and they'll return with official docs, real-world examples, and a warning about the footgun you're about to step on.

Role: External knowledge retrieval
Model: google/gemini-3-flash
Prompt: src/agents/librarian.ts

Documentation lookup, GitHub code search, library research, best practice retrieval. Read-only: they fetch wisdom; implementation is for others.

Designer

The Designer believes code should be beautiful - and so should everything it renders. Born from the frustration of a thousand ugly MVPs, they wield CSS like a brush and components like clay. Hand them a feature request; receive a masterpiece. They don't do "good enough."

Role: UI/UX implementation and visual excellence
Model: google/gemini-3-flash
Prompt: src/agents/designer.ts

Modern responsive design, CSS/Tailwind mastery, micro-animations, component architecture. Visual excellence over code perfection - beauty is the priority.

Tools & Capabilities

Tmux Integration

⚠️ Known Issue: When the server port is enabled, only one OpenCode instance can be opened at a time. We're tracking this in issue #15, and there's an upstream PR to OpenCode: opencode#9099.

Watch your agents work in real-time. When the Orchestrator launches sub-agents or initiates background tasks, new tmux panes automatically spawn showing each agent's live progress. No more waiting in the dark.

Why This Matters

What You Get

• Live Visibility: Each sub-agent gets its own pane showing real-time output
• Auto-Layout: Tmux automatically arranges panes using your preferred layout
• Auto-Cleanup: Panes close when agents finish, layout rebalances
• Zero Overhead: Works with OpenCode's built-in task tool AND our background_task tool

Quick Setup

1. Enable the OpenCode HTTP server in opencode.json (see OpenCode Config).
2. Enable tmux integration in oh-my-opencode-slim.json (see Plugin Config).
3. Run OpenCode inside tmux:

   tmux
   opencode

Layout Options

See the Option Reference for detailed configuration.

Quota Tool

For Antigravity users. You can trigger this at any time by asking the agent to "check my quota" or "show status."

Background Tasks

The plugin provides tools to manage asynchronous work:

LSP Tools

Language Server Protocol integration for code intelligence:

Code Search Tools

Fast code search and refactoring:

🧩 Skills

Skills are specialized capabilities that agents can use. Each agent has a default set of skills, which you can override in the agent config.

Available Skills

Default Skill Assignments

YAGNI Enforcement

The Minimalist's sacred truth: every line of code is a liability.

Use after major refactors or before finalizing PRs. Identifies unnecessary complexity, challenges premature abstractions, estimates LOC reduction, and enforces minimalism.

Playwright Integration

Browser automation for visual verification and testing.

• Browser Automation: Full Playwright capabilities (browsing, clicking, typing, scraping).
• Screenshots: Capture visual state of any web page.
• Sandboxed Output: Screenshots saved to session subdirectory (check tool output for path).

Customizing Agent Skills

Override skills per-agent in your Plugin Config:

{
  "agents": {
    "orchestrator": {
      "skills": ["*"]
    },
    "designer": {
      "skills": ["playwright"]
    }
  }
}

MCP Servers

Built-in Model Context Protocol servers (enabled by default):

Disabling MCPs

You can disable specific MCP servers by adding them to the disabled_mcps array in your Plugin Config.

Configuration

Files You Edit

Platform paths: On Windows, user config may also be at %APPDATA%\opencode\.

Plugin Config (oh-my-opencode-slim.json)

All plugin options in one file:

{
  "tmux": {
    "enabled": true,
    "layout": "main-vertical",
    "main_pane_size": 60
  },
  "disabled_agents": [],
  "disabled_mcps": ["websearch", "grep_app"],
  "agents": {
    "orchestrator": {
      "model": "openai/gpt-5.2-codex",
      "variant": "high",
      "skills": ["*"]
    },
    "explorer": {
      "model": "opencode/glm-4.7",
      "variant": "low"
    },
    "designer": {
      "skills": ["playwright"]
    }
  }
}

Option Reference

Uninstallation

1. Remove the plugin from your OpenCode config:

   Edit ~/.config/opencode/opencode.json and remove "oh-my-opencode-slim" from the plugin array.

2. Remove configuration files (optional):

   rm -f ~/.config/opencode/oh-my-opencode-slim.json
   rm -f .opencode/oh-my-opencode-slim.json

Credits

This is a slimmed-down fork of oh-my-opencode by @code-yeongyu.

License

MIT