# Agentic Development Environment (ADE) An Agentic Development Environment (ADE) is a development environment designed around AI coding agents rather than human-driven editors. Where traditional IDEs optimize for human code editing (syntax highlighting, autocomplete, debugging), ADEs optimize for orchestrating, reviewing, and controlling autonomous coding agents. ## Key Characteristics - **Agent orchestration**: manage multiple coding agents working in parallel on different tasks - **Isolation and sandboxing**: containerized workspaces with explicit disk and network controls so agents operate safely - **Bounded autonomy**: agents can work independently within defined boundaries, reducing constant approval friction - **Review-centric workflow**: the human role shifts from writing code to reviewing agent output (diffs, transcripts, artifacts) - **Multi-agent support**: a single interface for different agent backends ([[Claude Code]], [[Cursor.com|Cursor]], [[OpenAI Codex|Codex]], etc.) ## How It Differs from IDEs An IDE is built for a human typing code. An ADE is built for humans supervising agents that produce code. The core interaction loop changes from "write code → run → debug" to "define task → agent executes → review diff → merge." ## Examples - [[JetBrains Air]] — by [[JetBrains]]; public preview March 2026; multi-agent, multi-execution-mode (local/Docker/git worktree/cloud) - [[ctx (ADE)]] ## Related - [[JetBrains]] - [[JetBrains Air]] - [[IntelliJ IDEA]] - [[Agent Client Protocol (ACP)]] - [[Claude Code]] - [[Cursor.com|Cursor]] - [[Vibe Coding]] - [[AI Coding Maturity Levels]]