everything-opencode

Complete OpenCode configuration - agents, skills, hooks, commands, rules. Battle-tested patterns from everything-claude-code adapted for OpenCode.

Features

• 16 Specialized Agents - Planner, architect, code-reviewer, tdd-guide, security-reviewer, designer, multimodal-looker, fixer, and more
• 25 Slash Commands - /plan, /tdd, /e2e, /code-review, /build-fix, /refactor-clean, /design, /analyze-visual, etc.
• 24 Skills - Coding standards, TDD workflow, continuous learning, verification loops, frontend UI/UX, LSP validation
• 9 Rule Sets - Security, coding style, testing, git workflow, LSP validation guidelines
• Smart Hooks - Pre/post tool use hooks for tmux reminders, console.log warnings, TypeScript checks
• LSP Tools - Go-to-definition, find-references, diagnostics, rename
• Background Task Management - With optional tmux integration
• Flexible Model Configuration - Presets + per-agent overrides, mix providers freely

Installation

# Install the package
npm install -g everything-opencode

# Install agents, commands, skills, rules to your config
everything-opencode install

This copies all components to ~/.config/opencode/.

Configuration

Add to your opencode.json:

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

Model Configuration

Configure agent models in ~/.config/opencode/everything-opencode.json.

Quick Start: Use a Preset

{
  "preset": "github-copilot"
}

Available Presets (each uses only that provider's models):

| Preset | Provider | Models | |--------|----------|--------| | github-copilot | GitHub Copilot | claude-opus-4-5, claude-sonnet-4-5, claude-haiku-3-5 | | anthropic | Anthropic | claude-opus-4-5, claude-sonnet-4-5, claude-haiku-3-5 | | bedrock | AWS Bedrock | claude-opus-4-5, claude-sonnet-4-5, claude-haiku-3-5 | | google | Google | gemini-3-pro-high, gemini-3-flash | | openai | OpenAI | gpt-4o, o3, gpt-4o-mini |

Mix Providers with Agent Overrides

Use a preset as base, then override specific agents with different providers:

{
  "preset": "github-copilot",
  "agents": {
    "designer": { "model": "google/gemini-3-pro-high" },
    "multimodal-looker": { "model": "google/gemini-3-flash" }
  }
}

This gives you Claude via GitHub Copilot for most agents, but Gemini for vision/design tasks.

Override Any Agent

Each agent can be configured individually:

{
  "preset": "anthropic",
  "agents": {
    "orchestrator": { "model": "anthropic/claude-opus-4-5" },
    "doc-updater": { "model": "openai/gpt-4o-mini" },
    "designer": { "model": "google/gemini-3-pro-high" }
  }
}

Environment Variable Override

Switch presets without editing config:

EVERYTHING_OPENCODE_PRESET=bedrock opencode

Project-Level Overrides

Create .opencode/everything-opencode.json in your project root to override settings per-project:

{
  "agents": {
    "build-error-resolver": { "model": "anthropic/claude-opus-4-5" }
  }
}

Configuration Priority (highest to lowest):

1. Environment variable (EVERYTHING_OPENCODE_PRESET)
2. Project config (.opencode/everything-opencode.json)
3. User config (~/.config/opencode/everything-opencode.json)
4. Built-in preset defaults

Disable OpenCode Default Agents (Optional)

OpenCode includes default agents (plan, build, codebase) that overlap with ours (planner, build-error-resolver). To disable the defaults and use only our agents:

{
  "preset": "github-copilot",
  "disable_default_agents": true
}

This is optional - by default both sets of agents are available.

Tmux Integration (Optional)

Enable tmux for parallel session management:

{
  "preset": "github-copilot",
  "tmux": {
    "enabled": true,
    "layout": "main-vertical",
    "main_pane_size": 60
  }
}

What's Included

Agents (16)

| Agent | Purpose | |-------|---------| | orchestrator | Multi-agent coordination and task delegation | | planner | Implementation planning for complex features | | architect | System design and architectural decisions | | code-reviewer | Code quality and security review | | tdd-guide | Test-driven development guidance | | security-reviewer | Security vulnerability analysis | | build-error-resolver | Fix build and compilation errors | | e2e-runner | End-to-end test generation | | refactor-cleaner | Dead code removal and refactoring | | doc-updater | Documentation updates | | go-reviewer | Go-specific code review | | go-build-resolver | Go build error resolution | | database-reviewer | Database schema and query review | | designer | Frontend UI/UX specialist | | multimodal-looker | Vision analysis (screenshots, PDFs, images) | | fixer | Fast implementation for well-defined tasks |

Commands (25)

| Command | Description | |---------|-------------| | /plan | Create implementation plan | | /tdd | Test-driven development workflow | | /e2e | Generate E2E tests | | /code-review | Review code quality | | /build-fix | Fix build errors | | /refactor-clean | Clean up dead code | | /design | UI/UX design and implementation | | /analyze-visual | Analyze screenshots/images/PDFs | | /learn | Extract patterns mid-session | | /checkpoint | Save verification state | | /verify | Run verification loop | | /orchestrate | Multi-agent orchestration | | /eval | Evaluate session/code quality | | /evolve | Evolve learned patterns | | /go-build | Fix Go build errors | | /go-review | Go-specific code review | | /go-test | Go TDD workflow | | /instinct-export | Export learned patterns | | /instinct-import | Import learned patterns | | /instinct-status | Show continuous learning status | | /setup-pm | Setup project management | | /skill-create | Create new skill from template | | /test-coverage | Analyze test coverage | | /update-codemaps | Update code maps | | /update-docs | Update documentation |

Skills (24)

• coding-standards - Language best practices
• tdd-workflow - TDD methodology
• continuous-learning - Auto-extract patterns from sessions
• backend-patterns - API, database, caching patterns
• frontend-patterns - React, Next.js patterns
• frontend-ui-ux - Design principles and styling
• security-review - Security checklist
• golang-patterns - Go idioms and best practices
• lsp-validation - Type-safe code validation
• And more...

Hooks

Pre-Tool-Use:

• Tmux reminder for long-running commands
• Block dev servers outside tmux
• Block unnecessary markdown file creation
• Git push reminder

Post-Tool-Use:

• Console.log warning after file edits
• PR creation helper
• Optional TypeScript checking

Session Lifecycle:

• Context persistence across sessions

Compaction:

• Strategic compaction suggestions every 20 edits

CLI Commands

everything-opencode install   # Copy components to ~/.config/opencode/
everything-opencode version   # Show version
everything-opencode help      # Show help

Project Structure

everything-opencode/
├── agents/      # Markdown agent definitions
├── commands/    # Slash command definitions
├── skills/      # Skill directories with SKILL.md
├── rules/       # Always-follow guidelines
├── contexts/    # Dynamic context injection files
└── src/         # Plugin TypeScript code

Credits

Based on everything-claude-code by @affaanmustafa, adapted for OpenCode.

Background task management and LSP tools ported from oh-my-opencode-custom.

License

MIT