Research Note

Antigravity 2.0 Product Identity Record

Google's current overview defines Antigravity 2.0 as a standalone desktop application for managing agents across coding and knowledge work. It functions independently of

Aug 4, 20261 min readBy Dalton Anderson
In this article

Antigravity 2.0 Product Identity Record

Google's current overview defines Antigravity 2.0 as a standalone desktop application for managing agents across coding and knowledge work. It functions independently of an IDE.

Google's feature and project documentation describes projects, multiple folders, local and worktree modes, scoped settings, scoped permissions, artifacts, subagents, scheduled tasks, hooks, skills, MCP servers, and voice.

The Google Developers Blog describes Antigravity CLI as a Go-based terminal product that shares the Antigravity agent harness and supports asynchronous workflows. It also records a June 18, 2026 consumer transition away from service through Gemini CLI and Gemini Code Assist IDE extensions, while identifying different enterprise and API-key boundaries.

The Antigravity changelog shows continued separate entries for 2.0, CLI, SDK, and IDE, with gradual rollout. Do not assume feature parity or universal availability from a single announcement.

Sources

Follow the evidence.

  1. Antigravity changelogantigravity.google
  2. Antigravity 2.0 product pageantigravity.google
  3. Spotify episode recordpodcasters.spotify.com
  4. Antigravity MCP documentationantigravity.google
  5. Antigravity subagents documentationantigravity.google
  6. csrc.nist.gov: finalcsrc.nist.gov
  7. Antigravity permissions documentationantigravity.google
  8. Antigravity 2.0 overviewantigravity.google
  9. Antigravity 2.0 feature deep diveantigravity.google
  10. Antigravity hooks documentationantigravity.google
  11. Google people-first content guidancedevelopers.google.com
  12. Antigravity projects documentationantigravity.google
  13. Google Developers Blog: transitioning Gemini CLI to Antigravity CLIdevelopers.googleblog.com

From this episode

Two useful next steps.

Article · 1 min

How to Write a Verifiable AI Agent Goal

Define an autonomous agent goal with an outcome, context, constraints, verification, completion condition, authority boundary, and honest blocked-state behavior.

Research Note · 1 min

Verifiable Agent Goal Record

E118 says an agent goal should identify what to do, the conditions under which the result should be produced, and how to measure whether the work matched the intent.

Return to the episode