{"slug": "making-google-antigravity-work-like-codex-in-openclaw", "title": "Making Google Antigravity Work Like Codex in OpenClaw", "summary": "A developer built an open-source plugin that bridges Google Antigravity, a coding agent, with OpenClaw, an agent orchestration platform, enabling Antigravity to work inside OpenClaw the same way Codex does. The plugin uses a Python sidecar bridge that exposes an OpenAI-compatible endpoint, allowing OpenClaw to call Gemini models through Antigravity's harness. The project is available on GitHub and aims to prevent provider lock-in by letting any model be a first-class citizen in an agent workflow.", "body_md": "Google Antigravity is a powerful coding agent. OpenClaw is my agent orchestration platform. The problem: they don’t talk to each other.\n\nI wanted Antigravity to work inside OpenClaw the same way Codex does. Spawn a session, send a task, get a response. No manual copy-paste, no switching between tools. Just another model in the toolkit, callable like any other.\n\nSo I built a plugin that bridges them.\n\n## What It Does\n\nThe plugin registers Google Antigravity as a provider in OpenClaw. Once installed, you can use Gemini models through Antigravity’s harness the same way you’d use Claude through Codex or GPT through the OpenAI API.\n\n``` php\nOpenClaw provider plugin\n -> http://127.0.0.1:8080/v1/chat/completions\n -> Python bridge sidecar\n -> google-antigravity localharness\n -> Gemini / Antigravity runtime\n```\n\nThe architecture is a Python sidecar bridge that sits between OpenClaw’s Node.js process and Google’s `google-antigravity`\n\npackage. The sidecar exposes an OpenAI-compatible endpoint at `localhost:8080`\n\n, so OpenClaw talks to it the same way it talks to any other model provider.\n\n## Why a Sidecar?\n\nThe bridge runs as a separate process (macOS LaunchAgent) by design. Python dependencies, localharness lifecycle, crash recovery, and logs stay isolated from OpenClaw’s Node.js process. Tight coupling between a Node.js agent platform and a Python inference runtime would have been fragile.\n\nIf the sidecar crashes, OpenClaw keeps running. If OpenClaw restarts, the sidecar keeps its sessions. They communicate over HTTP on localhost, which is the simplest and most reliable interface.\n\n## What Works\n\n- OpenAI-compatible\n`/v1/chat/completions`\n\nendpoint - Non-streaming and SSE responses\n- Per-session harness isolation (concurrent agents don’t interfere)\n- OpenAI tool-call bridge\n- Models: gemini-3.5-flash, gemini-3.1-pro, gemini-3.1-flash\n\n## The Hard Parts\n\n**Session isolation.** Each OpenClaw session needs its own harness instance. If two agents share a session, their contexts bleed into each other. The bridge maintains strict session boundaries, even under concurrent load.\n\n**Tool-call translation.** Cloud APIs have standardized tool-call formats. Antigravity’s format is different. The bridge translates between OpenAI’s format and Antigravity’s native format, handling edge cases like partial outputs and retries.\n\n**Crash recovery.** A localharness process can die mid-task. The bridge detects this, restarts the process, and resumes without losing context. Cloud APIs handle this for you; locally, you have to build it yourself.\n\n## Open Source\n\nThe plugin is open source:\n\n[https://github.com/eonghk/openclaw-antigravity](https://github.com/eonghk/openclaw-antigravity)\n\nExperimental, macOS-first, designed for local OpenClaw installations. Install with:\n\n```\nnpm install -g openclaw-antigravity\nopenclaw-antigravity install\nopenclaw-antigravity doctor\n```\n\n## Why This Matters\n\nThe agent ecosystem is converging on a pattern: multiple model providers, each with different strengths, orchestrated by a single platform. Codex for one kind of task, Claude for another, Antigravity for a third.\n\nBut every provider has its own API, its own session model, its own tool-call format. Without bridges like this, you’re stuck switching between tools manually. With them, your agent platform becomes a true orchestration layer that can route tasks to the best model for the job.\n\nThat’s the goal. Not lock-in to one provider, but the ability to use any provider as a first-class citizen in your agent workflow.", "url": "https://wpnews.pro/news/making-google-antigravity-work-like-codex-in-openclaw", "canonical_source": "https://chenchen.guru/blog/building-openclaw-antigravity-plugin/", "published_at": "2026-07-26 05:07:17.328695+00:00", "updated_at": "2026-07-26 05:07:19.515554+00:00", "lang": "en", "topics": ["developer-tools", "ai-agents", "ai-tools", "artificial-intelligence"], "entities": ["Google Antigravity", "OpenClaw", "Codex", "Gemini", "OpenAI"], "alternates": {"html": "https://wpnews.pro/news/making-google-antigravity-work-like-codex-in-openclaw", "markdown": "https://wpnews.pro/news/making-google-antigravity-work-like-codex-in-openclaw.md", "text": "https://wpnews.pro/news/making-google-antigravity-work-like-codex-in-openclaw.txt", "jsonld": "https://wpnews.pro/news/making-google-antigravity-work-like-codex-in-openclaw.jsonld"}}