Home
原始文档链接 / Source
Google Antigravity
Google Antigravity is an agentic development platform, evolving the IDE into the agent-first era. Antigravity enables developers to operate at a higher, task-oriented level managing agents across workspaces, while retaining a familiar AI IDE experience at its core. Antigravity extracts agents into their own surface and provides them the tools needed to autonomously operate across the editor, terminal, and browser emphasizing verification and higher-level communication via tasks and artifacts. This capability enables agents to plan and execute more complex, end-to-end software tasks, elevating all aspects of development, from building features, UI iteration, and fixing bugs to research and generating reports.
Main Features
AI-powered IDE
An AI-powered IDE with all of the AI features that developers have come to rely on such as Agent, Tab, and Command.
Asynchronous Agents
Asynchronous, local agents that can work in parallel on all of your workspaces.
Agent Manager
New Agent Manager view for an agent-first experience built around planning mode, the conversation UI, and artifact review.
Multi-window
A multi-window product with an Editor, Manager, and Browser.
Browser Agent
Agent that can actuate the browser for you and to accomplish dev tasks like dashboard reads, SCM actions, UI testing, etc. in the Browser.
Core Surfaces
Editor
A fully-functional AI-powered IDE that maps to a single workspace.
Browser
(In preview) Browser-use agent capabilities to read & actuate on more surfaces beyond just the IDE.
Agent Manager
(In preview) An orchestration "no code" view to start and view tasks in a minimalist product focused on the conversation and artifacts.
Key Terms
- Agent: The primary AI modality within Antigravity. While the user can work tightly with an Agent within the Editor, they can also have multiple agents working across multiple codebases, orchestrated and monitored through the Agent Manager.
- Tab & Command: The other AI modalities within Antigravity, specifically within the text editor part of the editor surface. Tab is a more powerful "autocomplete" and Command is an inline instructive modality. From past experience, these do not get nearly as much use as the Agent.
- Artifacts: We define an artifact as anything that the agent creates to allow it to get its work done or communicate its accomplishments to the human user. These include rich markdown files, diff views, architecture diagrams, images, browser recordings, etc.