Home / Features

Features

Everything you need for AI-powered development workflows

Quick Navigation

Jump to any section to learn more

🤖

Agents

Intelligent agents for different tasks

→
🪝

Hooks

20+ built-in workflow automation hooks

→
🔌

MCPs

Model Context Protocol integration

→
💻

LSP Support

Full language server protocol support

→
🎯

Multi-Model

Orchestrate multiple AI models

→
🧪

Experimental

Early access to cutting-edge features

→

🤖 Intelligent Agents

Oh My OpenCode comes with powerful, specialized agents designed for different tasks:

Planner-Sisyphus

The default agent that provides intelligent planning and execution. Enabled by default, it helps break down complex tasks and execute them systematically.

Learn more →

Librarian

Specialized agent for documentation and code exploration. Perfect for finding relevant code, understanding codebases, and managing knowledge.

Learn more →

Explore

Code exploration agent that helps you navigate and understand large codebases efficiently.

Learn more →

Oracle

Question-answering agent that provides intelligent responses based on codebase context.

Learn more →

View All Agents →

🪝 Powerful Hooks System

20+ built-in hooks that automate and enhance your workflow:

  • todo-continuation-enforcer - Ensures tasks are completed
  • context-window-monitor - Monitors and manages context window usage
  • session-recovery - Automatically recovers from errors
  • session-notification - Notifies about session events
  • comment-checker - Validates code comments
  • grep-output-truncator - Truncates large grep outputs
  • tool-output-truncator - Manages tool output sizes
  • directory-agents-injector - Injects directory-specific agents
  • directory-readme-injector - Adds README context automatically
  • empty-task-response-detector - Detects empty responses
  • think-mode - Enables thinking mode for agents
  • anthropic-context-window-limit-recovery - Handles Anthropic limits
  • rules-injector - Injects custom rules
  • background-notification - Background process notifications
  • auto-update-checker - Checks for updates automatically
  • startup-toast - Shows startup notifications
  • keyword-detector - Detects keywords in conversations
  • agent-usage-reminder - Reminds about agent usage
  • non-interactive-env - Handles non-interactive environments
  • interactive-bash-session - Manages interactive sessions
  • empty-message-sanitizer - Cleans empty messages
  • compaction-context-injector - Manages context compaction
  • thinking-block-validator - Validates thinking blocks
  • claude-code-hooks - Claude-specific hooks
  • ralph-loop - Ralph agent loop management
  • preemptive-compaction - Preemptive context compaction

View Hooks Documentation →

🔌 MCP Integration

Model Context Protocol (MCP) support for enhanced capabilities:

Context7

Fetches up-to-date official documentation for libraries. Enabled by default, it ensures you always have access to the latest documentation.

Enabled by Default

grep.app

Ultra-fast code search across millions of public GitHub repositories. Find code examples, patterns, and solutions instantly.

Enabled by Default
Note: Both Context7 and grep.app MCP are enabled by default. You can disable them via configuration if needed.

Learn More About MCPs →

💻 LSP Support

Full Language Server Protocol support with refactoring tools:

  • Code Analysis - Full LSP analysis capabilities
  • Refactoring Tools - Rename, code actions, and more
  • Custom LSP Servers - Add your own LSP servers
  • Priority Management - Configure server priorities
  • Extension Mapping - Map servers to file extensions
  • Environment Configuration - Custom environment variables
  • Initialization Options - Customize server initialization

LSP Documentation →

🎯 Multi-Model Orchestration

Seamlessly work with multiple AI models to find the best tool for each task:

Model Comparison

Test which model has the sharpest logic, who's the debugging god, and who writes the best code.

Task-Specific Models

Use different models for different tasks - frontend, backend, debugging, prose writing.

Unified Interface

Switch between models seamlessly with a unified interface and configuration.

🧪 Experimental Features

Early access to cutting-edge features:

  • Preemptive Compaction - Automatically compacts context before hitting limits
  • Aggressive Truncation - More aggressive tool output truncation
  • Auto Resume - Automatically resumes sessions after recovery
  • DCP for Compaction - Dynamic Context Pruning for better compaction
Warning: Experimental features may change or be removed in future versions. Use with caution.

Experimental Features Guide →

Additional Capabilities

Session Recovery

Automatically recovers from errors, thinking block violations, and interruptions. Never lose your progress.

Context Window Management

Smart compaction and truncation to stay within token limits. Preemptive compaction prevents issues before they occur.

Directory-Specific Agents

Automatically inject agents and READMEs based on directory structure for better context.

Auto Updates

Automatic update checking with optional startup notifications. Stay up-to-date effortlessly.