1. How do I use AI Agent Studio?
  2. Overview of AI Studio CLI

Overview of AI Studio CLI

You can use the AI Studio Command Line Interface (CLI) to scaffold projects, fetch server artifacts on request, create and modify local artifact files, validate files, run workflow and application tests, inspect debugger sidecar state, and save or publish to remote services when explicitly requested. Use the CLI directly when the next step is known, the command is mechanical, and you can supply the required options confidently. Use a code assistant such as Codex or Claude Code with the CLI when the work requires planning, interpretation, multifile changes, workflow or application design, validation repair, or careful sequencing. The content in this chapter is mainly about using code assistant with AI Studio CLI.

Use Case Recommendation
User creating or redesigning a workflow or application from natural language, a screenshot, or a product specification Code Assistant + AI Studio CLI
User adding a workflow node, application panel, action, communication, template, or test where the exact implementation isn't obvious Code Assistant + AI Studio CLI
User fixing validation or test failures across related artifacts Code Assistant + AI Studio CLI
User who wants visual editing plus source files VS Code extension
User running a known create, inspect, validate, test, fetch, or save command AI Studio CLI directly
User making a narrow local edit with known flags AI Studio CLI directly

Direct CLI is recommended for these options:

  • Checking version, identity, or command help.
  • Validating a known file.
  • Listing or fetching a known server object when you explicitly want server state.
  • Running a focused workflow or app test suite.
  • Saving or publishing a reviewed local artifact when the release path says to do so.

A Code assistant is recommended for these options:

  • Deciding which artifact family and command sequence to use.
  • Creating a workflow or application from a business goal.
  • Choosing and wiring workflow nodes, applications, agents, widgets, actions, templates, and communications.
  • Coordinating related application-backed workflow stages such as InitDisplay, Query, InvokeAction, FillParameters, or SendCommunication.
  • Handling validation errors, test sync, local judge attachments, and report summaries.

We recommend using the CLI from the AI Studio project root without assuming a globally installed aistudio executable. Run the bundled script by using this path:

node .agents/skills/aistudio/scripts/aistudio.js <command> [options]