coding-agent

You can delegate coding tasks to AI agents like Codex, Claude Code, or Pi to automate code review, refactoring, and feature development. The agent runs within your git repository context to ensure safe execution.

Yes, you can run long-running coding agent tasks in the background. This allows you to offload processes like refactoring large legacy codebases while continuing other work in your environment.

You can use interactive terminal execution by enabling pseudo-terminal support, setting pty to true. This allows interactive command-line interfaces to function properly when controlling AI coding agents.

Yes, the coding agent requires a git repository context for security and functionality. Running agents within a git repository ensures safe execution and proper tracking of code modifications.

When delegating code generation to Codex, specific execution modes like --full-auto and --yolo are supported. These modes allow automated processing of coding tasks without continuous manual intervention.

You can spawn an AI agent in a specific project directory by setting the workdir parameter. This directs the coding agent to operate within that exact environment for targeted iterative development.