opencode

This skill acts as a comprehensive technical companion for developers utilizing the OpenCode.ai ecosystem. It encapsulates the full breadth of the OpenCode command-line interface and Terminal User Interface (TUI), providing deep operational knowledge for AI-assisted development workflows. Whether you are performing initial project setup, managing complex agent configurations, or optimizing your development environment via the .opencode directory, this skill ensures accurate, high-fidelity guidance based on official documentation and industry best practices.

The skill provides critical support for managing AI agent behavior, including custom command creation, model switching, and tool integration. It is designed to assist software engineers in navigating the complexities of modern terminal-based AI tools, offering precise instructions on installation, auth connectivity, session management, and file-based context injection. It covers everything from basic operational commands to advanced configuration techniques found in AGENTS.md and opencode.json files.

• Full coverage of OpenCode CLI commands including auth, sessions, model management, and non-interactive execution modes.

• Detailed instructions for TUI navigation, leader key shortcuts, build vs. plan mode switching, and subagent invocation.

• Comprehensive configuration assistance for .opencode/opencode.json, covering agent permissions, tool permissions, and model parameters.

• Guidance on custom command templates using YAML frontmatter, shell injection patterns, and argument passing ($ARGUMENTS, $1).

• Deep integration support for file references (@), bash command execution (!), and git-based undo/redo operations.

• Practical tips for Claude-compatible configuration, reading CLAUDE.md files, and managing agent-specific settings.

• Troubleshooting and best practices for project initialization (/init), connection management (/connect), and multi-agent orchestration.

• Users should trigger this skill when initiating new projects, modifying existing agent configurations, or needing syntax reminders for custom commands.

• Expected inputs range from specific error codes and command flags to general questions about project architecture and TUI functionality.

• Constraints include platform-specific installation methods (npm, Homebrew, curl scripts) and local configuration file requirements (global vs. project-level).

• Designed for power users and developers seeking to maximize their efficiency in terminal environments via LLM-driven automation.