How do I switch between Build and Plan agents?

You can toggle between the Build (full access) and Plan (read-only) agents by pressing the Tab key or using the --agent flag in the CLI.

Does OpenCode support Model Context Protocol (MCP)?

Yes, the OpenCode Expert skill provides guidance on adding, listing, and debugging MCP servers to extend the agent's tool capabilities.

Can I undo changes made by the OpenCode agent?

Yes, OpenCode supports a stackable undo system via the /undo command, allowing you to revert and redo recent file edits easily.

OpenCode is an open-source AI coding agent designed for terminal-based development, featuring an interactive TUI, headless server modes, and customizable agents.

OpenCode Expert

Name: OpenCode Expert
Author: s-hiraoku

bys-hiraoku

开发者工具

Provides comprehensive guidance and command references for OpenCode, the open-source AI coding agent for terminal-based development.

关于

The OpenCode Expert skill empowers developers to master the OpenCode ecosystem, an open-source alternative for AI-assisted terminal coding. It offers instant access to TUI navigation, CLI command references, and keyboard shortcuts, while helping users manage specialized agent behaviors like 'Build' and 'Plan' modes. Whether you are configuring Model Context Protocol (MCP) servers, initializing project-specific agents, or utilizing advanced tools for file editing and shell execution, this skill provides the domain-specific knowledge needed to maximize productivity within the OpenCode environment.

主要功能

Configuration support for MCP servers and subagents
0 GitHub stars
Dual-agent workflow guidance for Build and Plan modes
Slash command integration for session and file management
Comprehensive CLI and TUI command reference
Detailed documentation for built-in tools like bash, lsp, and patch

使用场景

Executing and reverting AI-driven project refactors with the Build agent and undo stack
Navigating and analyzing complex codebases using @explore subagents and fuzzy file search
Setting up headless API servers and web interfaces for remote AI coding sessions

关于

主要功能

Configuration support for MCP servers and subagents
0 GitHub stars
Dual-agent workflow guidance for Build and Plan modes
Slash command integration for session and file management
Comprehensive CLI and TUI command reference
Detailed documentation for built-in tools like bash, lsp, and patch

使用场景

Executing and reverting AI-driven project refactors with the Build agent and undo stack
Navigating and analyzing complex codebases using @explore subagents and fuzzy file search
Setting up headless API servers and web interfaces for remote AI coding sessions