Code Agent FAQs

Question 1

How do file references work in Code Agent?

Accepted Answer

Code Agent uses a simple '@' syntax (e.g., @src/main.js) to automatically inject local file content into the AI context. This ensures the model has precise information about your existing codebase without manual copying and pasting.

Question 2

When should I use Code Agent over standard AI tools?

Accepted Answer

You should use Code Agent for large-scale refactoring across multiple files, complex code analysis requiring deep context, and automated workflows that benefit from running tasks in parallel or switching backends based on specific model strengths.

Question 3

Which AI backend should I choose for my task?

Accepted Answer

Codex is best for deep code analysis and complex logic; Claude excels at technical documentation and prompt engineering; and Gemini is optimized for UI/UX prototyping and layout generation. You can even chain different backends together in a single parallel workflow.

Question 4

Does Code Agent support long-running or multi-step tasks?

Accepted Answer

Yes, Code Agent includes session resumption capabilities. Every interaction generates a unique Session ID, allowing you to follow up on previous tasks or resume complex, multi-step AI interactions at any time.

Question 5

What does the Code Agent skill do?

Accepted Answer

Code Agent is a powerful orchestrator that allows you to execute complex coding tasks across multiple AI models including Codex, Claude, and Gemini. it supports advanced features like parallel task execution, local file references via @syntax, and persistent session management for multi-step workflows.

Code Agent

Code Agent

关于

主要功能

使用场景

关于

主要功能

使用场景