OpenCode Plugin Ecosystem

OpenCode achieves powerful extensibility through its plugin system. This guide covers six core plugins that solve multi-model orchestration, context management, web search, code editing, knowledge learning, and authentication challenges.

Core Plugins Overview

Plugin	Purpose	Source	Key Features
oh-my-opencode	Multi-model orchestration, sub-agent execution, autonomous workflows	npm	Professional agent orchestration
@tarquinen/opencode-dcp	Auto-compress redundant context, save tokens	npm	Intelligent context compression
opencode-websearch-cited	Web search with Google Search API	npm	Real-time information retrieval
morph-fast-apply	Ultra-fast code editing (10,500+ tokens/s, 98% accuracy)	GitHub	High-speed code application
smart-codebase	AI learns from sessions and code, generates SKILL files	npm	Continuous knowledge accumulation
opencode-openai-codex-auth	OpenAI OAuth authentication	npm	Secure authentication

---

1. oh-my-opencode

Detailed installation and configuration guide: oh-my-opencode Installation Guide

Core Capabilities

• Multi-Agent Architecture: Built-in agents like Explore, Oracle, Librarian, Sisyphus for specialized tasks
• Category Routing: Automatically selects optimal model based on task type
• Ultrawork Mode: /ulw keyword triggers autonomous work
• Prometheus Planning: Automatic decomposition, planning, and execution for complex projects
• Skills System: Built-in skills like playwright, frontend-ui-ux, git-master

Installation

npm install oh-my-opencode@latest
# or
bun add oh-my-opencode

---

2. @tarquinen/opencode-dcp

Overview

DCP (Dynamic Context Pruning) is an intelligent context management plugin that optimizes token usage through automatic compression and redundancy elimination.

Core Capabilities

• Automatic Context Compression: Identifies and compresses repetitive or redundant conversation history
• Smart Key Information Retention: Preserves context critical to current tasks
• Token Optimization: Significantly reduces API call costs
• Transparent Operation: Works automatically in the background without manual intervention

Installation

cd ~/.config/opencode
npm install @tarquinen/opencode-dcp@latest

Use Cases

• Long Conversations: Auto-compresses when conversation history exceeds thousands of tokens
• Multi-Round Iterations: Saves costs in tasks requiring multiple interactions like code refactoring
• Large Projects: Maintains performance when handling projects with extensive file context

How It Works

Configuration Example

DCP typically requires no additional configuration and auto-enables after installation. For custom behavior, add to opencode.json:

{
  "plugins": {
    "@tarquinen/opencode-dcp": {
      "enabled": true,
      "compressionThreshold": 8000,
      "preserveRecentMessages": 10
    }
  }
}

---

3. opencode-websearch-cited

Overview

Websearch Cited provides real-time web search capabilities for OpenCode, fetching latest information via Google Search API with cited sources.

Core Capabilities

• Real-Time Information Retrieval: Access to latest web information
• Source Citation: Every information point includes verifiable source links
• Intelligent Summarization: Automatically extracts and summarizes search results
• Multi-Language Support: Supports Chinese, English, and other languages

Installation

cd ~/.config/opencode
npm install opencode-websearch-cited@latest

Configuration Requirements

Requires Google Search API credentials:

# Method 1: Environment variables
export GOOGLE_SEARCH_API_KEY="your-api-key"
export GOOGLE_SEARCH_ENGINE_ID="your-engine-id"

# Method 2: Write to .env file
echo "GOOGLE_SEARCH_API_KEY=your-api-key" >> ~/.config/opencode/.env
echo "GOOGLE_SEARCH_ENGINE_ID=your-engine-id" >> ~/.config/opencode/.env

Use Cases

Scenario 1: Technical Queries

User: Search for React 19 latest features

Agent: [Calling websearch-cited]
Based on search results, React 19 introduces:
1. **React Compiler**: Automatic component performance optimization [Source: react.dev]
2. **Actions**: Simplified form handling and data submission [Source: github.com/facebook/react]
3. **Document Metadata**: Native support for <title> and <meta> tags [Source: react.dev/blog]

Scenario 2: Market Research

User: Find 2025 AI coding tools market trends

Agent: [Calling websearch-cited]
Current AI coding tools market shows:
- GitHub Copilot leads market share [Source: stackoverflow.com/survey]
- Cursor and Windsurf growing rapidly [Source: techcrunch.com]
- Enterprise adoption up 40% [Source: gartner.com]

Workflow

Best Practices

1. Clear Search Intent: Use specific keywords rather than vague descriptions
2. Verify Sources: Check authority of cited links
3. Consider Timeliness: For rapidly changing tech, prioritize recent results
4. Combine Knowledge: Merge search results with local knowledge for comprehensive answers

---

4. morph-fast-apply

Overview

Morph Fast Apply is a revolutionary code editing plugin achieving ultra-fast code application through Morph LLM's dedicated API, reaching 10,500+ tokens/s with 98% accuracy.

Core Capabilities

• Ultra-Fast Editing: 10-50x faster than traditional methods
• High Accuracy: 98% code application success rate
• Intelligent Diff: Precisely identifies code segments needing modification
• Batch Operations: Supports simultaneous multi-file editing
• Conflict Detection: Automatically detects and alerts potential code conflicts

Technical Advantages

Feature	Traditional Editing	Morph Fast Apply
Speed	200-500 tokens/s	10,500+ tokens/s
Accuracy	85-90%	98%
Multi-File Support	Limited	Native
Conflict Detection	Manual	Automatic
Large File Handling	Difficult	Optimized

Installation

# Clone plugin repository
mkdir -p ~/.config/opencode/plugins
git clone https://github.com/JRedeker/opencode-morph-fast-apply ~/.config/opencode/plugins/morph-fast-apply

# Install dependencies
cd ~/.config/opencode/plugins/morph-fast-apply
npm install

Configure API Key

Visit Morph Dashboard to obtain API Key:

# Method 1: Environment variable (recommended)
# macOS/Linux
echo 'export MORPH_API_KEY="your-morph-api-key"' >> ~/.bashrc  # or ~/.zshrc
source ~/.bashrc

# Windows PowerShell (Administrator)
[System.Environment]::SetEnvironmentVariable("MORPH_API_KEY", "your-key", "User")

# Method 2: Write to .env file
echo "MORPH_API_KEY=your-morph-api-key-here" >> ~/.config/opencode/.env

Use Cases

Scenario 1: Large-Scale Refactoring

User: Refactor all class components to function components in the project

Agent: [Using morph-fast-apply]
✓ Analyzed 45 component files
✓ Identified 23 class components
✓ Batch converting...
✓ Complete! Average 0.3 seconds per file

Scenario 2: API Upgrade

User: Upgrade all React Router v5 code to v6 syntax

Agent: [Using morph-fast-apply]
✓ Detected 18 route files
✓ Convert <Switch> to <Routes>
✓ Update <Route> component syntax
✓ Change useHistory to useNavigate
✓ All files updated, no conflicts

Scenario 3: Code Style Unification

User: Unify import statement order across project (React -> Third-party -> Local)

Agent: [Using morph-fast-apply]
✓ Scanned 127 files
✓ Reordered imports in 89 files
✓ Preserved all functionality
✓ Completed in 8 seconds

How It Works

Performance Comparison

Real test data (based on 50-file refactoring task):

Traditional Method:
- Total time: ~5 minutes
- Success rate: 87%
- Manual fixes needed: 6 files

Morph Fast Apply:
- Total time: ~15 seconds
- Success rate: 98%
- Manual fixes needed: 1 file

Best Practices

1. Test Small First: Test on small scope before large-scale application
2. Use Version Control: Ensure all changes are under Git management for easy rollback
3. Batch Processing: For huge projects (1000+ files), process in batches
4. Verify Build: Run tests and build immediately after applying changes
5. Review Diff: Manually review diffs for critical code modifications

---

5. smart-codebase

Overview

Smart-Codebase is a knowledge accumulation plugin that enables OpenCode to learn from every session and codebase, automatically generating SKILL files. New sessions automatically load project context without repetitive explanations.

Core Capabilities

• Auto Learning: Extracts knowledge from sessions, automatically writes SKILL files
• sc-init: Scans source code, generates project skill files
• sc-update: Analyzes current session, updates skill library
• Structured Storage: SKILL.md + reference/*.md layered organization
• Cross-Session Persistence: Knowledge saved in files, auto-loaded in next session

Installation

cd ~/.config/opencode
npm install smart-codebase@latest

# or using bun (recommended)
bun add smart-codebase

Usage

# Run slash commands in OpenCode
/sc-init      # Initialize project skills
/sc-update    # Extract knowledge from current session

File Structure

~/.config/opencode/skills/
└── <project-name>/
    ├── SKILL.md           # Main skill file (auto-discovered)
    └── reference/
        ├── architecture.md  # Architecture documentation
        ├── patterns.md      # Pattern documentation
        └── gotchas.md       # Pitfall guide

Workflow

Best Practices

1. First Project: Run /sc-init to establish baseline
2. Milestone Moments: Run /sc-update to save important decisions
3. Knowledge Layering: Summary in SKILL.md, details in reference/
4. Manual Review: Regularly review generated skill files

---

6. opencode-openai-codex-auth

Overview

OpenAI Codex Auth provides OpenAI OAuth authentication support for OpenCode's GPT model calls.

Installation

cd ~/.config/opencode
npm install opencode-openai-codex-auth

Configuration

After installation, run:

opencode providers

Select OpenAI provider and complete OAuth authentication flow.

---

Plugin Synergy

These six plugins work together seamlessly to form powerful workflows:

Typical Workflow Example

Real-World Case: Complex Project Refactoring

User: Refactor my e-commerce backend, referencing latest microservice best practices

Step 1: [oh-my-opencode Planning]
→ Prometheus analyzes project structure
→ Generate refactoring plan

Step 2: [websearch-cited Search]
→ Get microservice architecture best practices
→ Get relevant technical documentation

Step 3: [DCP Compression]
→ Compress search results
→ Retain key migration steps

Step 4: [morph-fast-apply Execution]
→ Batch update code
→ Preserve functionality

Step 5: [smart-codebase Learning]
→ /sc-update saves decisions
→ Next session auto-loads

Result: Complete tech stack refactoring in < 10 minutes

---

Plugin Comparison & Selection

Feature Matrix

Use Case	Recommended Plugin	Reason
Multi-model orchestration/sub-agents	oh-my-opencode	Professional agent orchestration
Long conversation optimization	DCP	Auto-manages context, reduces costs
Need latest information	websearch-cited	Real-time web search
Large-scale code changes	morph-fast-apply	Fast with high accuracy
Knowledge accumulation/learning	smart-codebase	Auto-learn from sessions
Technical research	websearch-cited	Authoritative sources
Code refactoring	morph-fast-apply	Strong batch processing
Token cost control	DCP	Intelligent redundancy compression
OpenAI authentication	openai-codex-auth	OAuth secure auth
Autonomous work	oh-my-opencode	Ultrawork mode

Cost-Benefit Analysis

Scenario: Handle refactoring task with 100 files

Without Plugins:
- Time cost: ~2 hours
- Token cost: ~500K tokens ($10)
- Manual fixes: ~30 minutes

With Plugin Combo:
- Time cost: ~5 minutes
- Token cost: ~50K tokens ($1)
- Manual fixes: ~5 minutes

Savings: 95% time + 90% cost

---

Complete Installation & Configuration Guide

One-Click Install Script

#!/bin/bash
# install-opencode-plugins.sh

echo "🚀 Starting OpenCode core plugins installation..."

# Enter config directory
cd ~/.config/opencode || exit

# Install npm plugins
echo "📦 Installing npm plugins..."
npm install @tarquinen/opencode-dcp@latest \
            opencode-websearch-cited@latest \
            smart-codebase@latest

# Install morph-fast-apply
echo "📦 Installing morph-fast-apply..."
mkdir -p plugins
git clone https://github.com/JRedeker/opencode-morph-fast-apply plugins/morph-fast-apply
cd plugins/morph-fast-apply
npm install

# oh-my-opencode detailed config in ./oh-my-opencode-installation-guide
npm install oh-my-opencode@latest

# Symlink smart-codebase to plugins (if needed)
ln -sf ../node_modules/smart-codebase plugins/smart-codebase

echo "✅ All plugins installed!"
echo "⚠️  Please configure these environment variables:"
echo "   - MORPH_API_KEY (required for morph-fast-apply)"
echo "   - GOOGLE_SEARCH_API_KEY (optional for websearch)"
echo "   - GOOGLE_SEARCH_ENGINE_ID (optional for websearch)"

Verify Installation

# Check plugin loading status
opencode debug config | grep -A 10 "plugins"

# Test DCP
opencode run "Long conversation test"

# Test websearch-cited
opencode run "Search latest AI news"

# Test morph-fast-apply
opencode run "Create a simple React component"

# Initialize smart-codebase (first time)
opencode run "/sc-init"

---

FAQ

Q1: Will DCP lose important context?

A: No. DCP uses intelligent algorithms to identify key information, only compressing truly redundant content. If concerned, adjust the preserveRecentMessages parameter.

Q2: Does websearch-cited require payment?

A: Google Search API has free quota (100 queries/day), payment required beyond that. Usually sufficient for personal use.

Q3: What's the API cost for morph-fast-apply?

A: Morph API charges by token, but due to extreme speed, actual cost is typically lower than using general LLM multi-turn conversations.

Q4: Which models do these plugins support?

A:

• DCP, websearch-cited, smart-codebase: Support all OpenCode-compatible models
• morph-fast-apply: Uses dedicated Morph API, independent of main model
• oh-my-opencode: Selects different models based on configuration (GPT, Claude, Gemini, etc.)

Q5: Can I install only some plugins?

A: Yes. These six plugins are completely independent and can be selectively installed based on needs.

Q6: What's the difference between smart-codebase and oh-my-opencode's knowledge systems?

A:

• smart-codebase: Extracts knowledge from current session, automatically generates SKILL files. Suitable for project-level knowledge accumulation.
• oh-my-opencode: Built-in skills system with preset skills like playwright, frontend-ui-ux. Suitable for cross-project reuse.

The two can be used together: oh-my-opencode provides base skills, smart-codebase supplements project-specific knowledge.

---

Advanced Tips

Tip 1: Customize DCP Compression Strategy

{
  "plugins": {
    "@tarquinen/opencode-dcp": {
      "compressionThreshold": 6000,
      "preserveRecentMessages": 15,
      "aggressiveMode": false,
      "preserveCodeBlocks": true
    }
  }
}

Tip 2: Configure Websearch Preferences

{
  "plugins": {
    "opencode-websearch-cited": {
      "maxResults": 10,
      "language": "en-US",
      "safeSearch": "moderate",
      "dateRestrict": "m1"  // Only search last month
    }
  }
}

Tip 3: Optimize morph-fast-apply Performance

{
  "plugins": {
    "morph-fast-apply": {
      "maxConcurrentFiles": 10,
      "timeout": 30000,
      "retryOnFailure": true,
      "validateSyntax": true
    }
  }
}

---

Summary

These six core plugins represent the powerful extensibility of the OpenCode ecosystem:

• oh-my-opencode: Multi-model orchestration, sub-agent execution, Ultrawork autonomous work
• @tarquinen/opencode-dcp: Makes long conversations more economical and efficient
• opencode-websearch-cited: Connects to real-time internet knowledge
• morph-fast-apply: Elevates code editing speed to new heights
• smart-codebase: Enables AI to continuously learn from sessions
• opencode-openai-codex-auth: Secure OAuth authentication

By properly combining these plugins, you can build extremely efficient AI-assisted development workflows, dramatically reducing time and cost investment while maintaining high-quality output.

---

Related Resources

• OpenCode Official Documentation
• oh-my-opencode Detailed Guide
• Morph LLM Dashboard
• Google Custom Search API
• OpenCode Skills Ecosystem