Getting started with Google Antigravity

Google Antigravity transforms your regular IDE into an agentic development platform. Beyond basic tab completion, Antigravity offers an intuitive, artifact-based approach to monitoring and providing feedback to agents. By utilizing rules, workflows, and skills, you can fully customize the environment to your team’s best practices, while a dedicated browser subagent handles tasks outside the editor. Last but not least, its Agent Manager can spin up and coordinate multiple parallel agents across various workflows, allowing you to work at a higher level than the code itself.

Transcript

1. Getting Started with https://antigravity.google
2. Who am I? Mete Atamel Developer Advocate @ Google Cloud atamel.dev speakerdeck.com/meteatamel @meteatamel
3. What is Antigravity? An agentic development platform from Google allowing anyone to build
4. Antigravity Platform  Antigravity (2.0)  Antigravity IDE & Extensions  Antigravity CLI  Antigravity SDK
5. Antigravity 2.0
6. Antigravity 2.0 Your command center to manage multiple local agents in parallel Group conversations into projects and operate across multiple workspaces
9. Projects A project defines the filesystem and security boundaries for the agent Multi-folder & worktree support, scoped settings and permissions that the agent respects
10. Review and Terminal You can review generated code but no inline editing There’s a basic terminal
11. Scheduled Tasks Automate routine tasks with Scheduled Tasks Define a cron schedule and the agents start and run autonomously in the background
12. Slash commands /goal: Run until the task is finished /grill-me: Interview me to align on a plan /learn: Distills into persistent Rules or Skills /schedule: Schedule a task /browser: Use the browser subagent /btw: Asks a question in the background
13. Subagents Subagents are instantiated dynamically to tackle parallel parts of complex problems
14. Subagents Pre-packaged with some specialized subagents: research, browser, self You can define and use your own custom agents
15. Browser Subagent Antigravity has the ability to open and control a Chrome browser Requires Chrome and a permission to start a debugging session
16. Antigravity IDE Antigravity IDE Extensions
17. Antigravity IDE A VS Code-based AI-assisted editor with an agent side panel and terminal Auto-complete, tab to import, tab to jump, commands, explain & fix, send problems to agent, send terminal output to agent
18. AI-powered IDE Agent Terminal
19. Models
20. Auto complete
21. Tab to import
22. Tab to jump
23. Explain & fix
24. Send (problems) to agent
25. Send terminal output to agent
26. Antigravity IDE Extensions IDE Extensions with Antigravity agent side Panel (but no AI-assistance in the editor) Extensions for: Visual Studio Code, Visual Studio, JetBrains, Zed, XCode
27. Antigravity IDE Extensions
28. Antigravity IDE Extensions
29. Antigravity CLI
30. Antigravity CLI The lightweight, fast, terminal-first surface to work with Antigravity agents
31. Antigravity SDK
32. Antigravity SDK The Antigravity SDK gives you the same tools, agent loop, and context management that power Antigravity, programmable in Python
33. Antigravity SDK import asyncio from google.antigravity import Agent, LocalAgentConfig async def main(): config = LocalAgentConfig() async with Agent(config) as agent: response = await agent.chat("What files are in the current directory?") print(await response.text()) if __name__ == "__main__": asyncio.run(main())
35. Artifacts Task List Implementation Plan Code Diff Walkthrough Screenshot & Recording
37. Task List Used by the agent to keep track and monitor progress with action items Typically, you do not need to directly interact with the task list
38. Task List
39. Implementation Plan Agent creates an implementation plan for technical changes within your codebase Meant to be reviewed by the user, unless Always Proceed is enabled
40. Implementation Plan
41. Code Diff As Antigravity works on a task, it’ll create code diffs for you to accept or reject
42. Code Diff
43. Walkthrough Agent creates a walkthrough after a task is done Summary of the code changes and steps taken (screenshots, browser recordings) to verify the changes
44. Walkthrough
45. Screenshot & Recording Agent can take browser screenshots and recordings to verify functionality and add it to the walkthrough
46. Screenshot & Recording
47. Customizations Skills MCP Servers Rules & Workflows Hooks Sidecars Custom Agents Plugins
48. Skills An open standard for extending agent capabilities with best practices During a conversation, the agent sees a list of available skills and decides which one to use for the task
49. Define a Skill my-skill/ ├─── SKILL.md ├─── scripts/ ├─── examples/ └─── resources/ # Main instructions (required) # Helper scripts (optional) # Reference implementations (optional) # Templates and other assets (optional)
50. MCP Servers Antigravity supports Model Context Protocol (MCP) servers Provides the AI with real-time context beyond just the files open in your editor
51. MCP Servers
52. Rules Rules help guide the behavior of the agent For example, you can add a rule for agent to follow a certain code style, or to always document methods
53. Workflows Workflows are saved prompts that you can trigger on demand with / with the agent For example, you can add a workflow to generate unit tests on demand
54. Rules and Workflows
55. Rules vs. Workflows vs. Skills Rules ≅ system instructions (always-on) Workflows ≅ saved prompts (user-triggered) Skills ≅ saved prompts, scripts, examples (agent-triggered)
56. Hooks
57. Sidecars
58. Custom Agents
59. Plugins
60. Where does Antigravity look for configurations? atamel.dev/posts/2026/08-21_where_agy_configuration_summary/
61. Permissions
62. Hierarchy of permissions Global (User) Level The highest layer. Establishes baseline security policies, default safe directories, and absolute restrictions across all workspaces. Project Level Inherits and refines global policies. Configures settings for specific repositories, workspace-level permissions, and local trusted tools. Conversation (Session) Level The most granular layer. Dynamically manages individual execution sessions, temporary folder access, and on-the-fly execution approvals.
63. Permissions Outside of Folders Command Auto Execution ✓ Always Ask DEFAULT ✓ Require Review DEFAULT Allow Proceed in Sandbox Deny Always Proceed Enable Sandbox Mode Artifact Review Policy ✓ False DEFAULT ✓ Always Ask DEFAULT True Always Proceed
64. Permissions: Allow, Ask, Deny File Access Rules Ask Allow Deny Network Access Rules Terminal Commands Commands Outside Sandbox Ask Ask Ask Allow Allow Allow Deny Deny Deny MCP Tools Ask Allow Deny
65. Permissions Browser Javascript Execution Policy Disable ✓ Request Review DEFAULT Always Proceed Browser Actuation Rules (Execute URLs) Allow Ask Deny
66. Permission format
67. Login Options & Plans
68. Login Options In Antigravity, you are presented with two primary authentication routes: personal or business (Google Cloud)
69. Personal Plan “Continue with Google” will use your gmail account and offer either the Starter Quota plan (Free Tier) or the Pro or Ultra plans you may have signed up for, which will improve your quota
70. Business (Google Cloud) Plan “Use business account” offers either a pay-as-you-go model or the use of a Gemini Enterprise license with a method of payment linked to your Google Cloud environment
72. Remote Control Remote control allows you to connect and drive your Antigravity sessions running across your different machines with a web browser
73. Antigravity codelabs codelabs.developers.google.com/?product=antigravity
74. Thank You! Mete Atamel Developer Advocate @ Google Cloud atamel.dev speakerdeck.com/meteatamel @meteatamel