Features

Escape AI features — everything the agent actually does

No invented capabilities — this is what ships today, across chat, desktop, and the VS Code extension.

✎

Reviewable diffs

Every change the agent proposes comes back as a diff, file by file. You accept, reject, or edit it before anything touches your repo — nothing lands silently.

⌘

Terminal & command execution

The agent can run builds, tests, and installs on your behalf — always with your explicit permission first, never running commands silently in the background.

⎇

Native git integration

Branches and commits happen where your code actually lives, using your existing git setup — no separate versioning system to learn.

▣

MCP server support

Connect any MCP-compatible server by giving it a name, command, and arguments in Settings — filesystem, database, or your own internal tools. There's no curated one-click directory; you wire up what you need.

◈

Multi-model routing

OpenAI powers chat and knowledge retrieval by default, including gpt-6-astra for Pro, Team, and Enterprise plans. Gemini is available as an opt-in provider that teams can configure when they want it.

⇶

Parallel Agent Tasks

Kick off several background tasks at once and review each one independently as it finishes, instead of waiting on a single agent run at a time.

See it running in your own repo

Free to start. No credit card required.