Agent OS is designed to work with any AI coding tool. Whether you use Claude Code, Cursor, Codex, Gemini, Windsurf, or whatever hot new tool drops next week, you can use Agent OS with it.
Agent OS adapts to your tool through flexible configuration options that let you enable the features your tool supports:
- Using with Claude Code - Full integration with commands, subagents, and Skills
- Using with other tools - Sequential prompt-based workflow for any AI coding tool
Using Agent OS with Claude Code
Claude Code provides the most integrated experience with Agent OS through native support for slash commands, subagents, and Skills.
When configured for Claude Code (claude_code_commands: true), Agent OS:
- Generates slash commands in your
.claude/commands/agent-os/folder (like/shape-spec,/write-spec,/create-tasks) - Optionally converts your standards into Claude Code Skills for intelligent, context-efficient standards application
- Enables orchestration with specialized subagents when
use_claude_code_subagents: true - Provides the complete spec-driven development workflow through easy-to-use slash commands
Learn more about Claude Code configuration options
Using Agent OS with other AI coding tools
Agent OS works with any AI coding tool through a sequential prompt-based workflow. When configured for other tools (agent_os_commands: true), Agent OS generates markdown command files that you can reference in your tool's chat interface.
When configured for other AI tools, Agent OS:
- Generates workflow command files in your
agent-os/commands/folder - Injects your standards and workflows into each command
- Provides clear, step-by-step prompts for the complete development workflow
Running commands in other tools
Agent OS provides two ways to run each workflow command with other AI tools:
Option A: Run the main command file, which instructs your agent to complete all steps automatically. For example:
@agent-os/commands/write-spec/write-spec.md run thisOption B: Run each numbered command file sequentially for more control and reliability. For example:
@agent-os/commands/write-spec/1-[step-name].md run this
@agent-os/commands/write-spec/2-[step-name].md run thisSome agents perform better with Option A, while others work better with the step-by-step approach of Option B.
Follow the Agent OS workflow to see the complete sequence of commands for building features with your AI tool. Each workflow page shows both options.