๐ ๐ ๐ฅ๐ฎ๐ฏ๐ฏ๐ถ๐ ๐ฅ๐ญ ๐ต๐ฎ๐ฑ ๐ฏ๐ฒ๐ฐ๐ผ๐บ๐ฒ ๐ฎ ๐ฑ๐ฟ๐ฎ๐๐ฒ๐ฟ ๐ผ๐ฟ๐ป๐ฎ๐บ๐ฒ๐ป๐. ๐ง๐ต๐ถ๐ ๐๐ฒ๐ฒ๐ธ๐ฒ๐ป๐ฑ ๐ ๐ด๐ฎ๐๐ฒ ๐ถ๐ ๐ฎ ๐ฟ๐ฒ๐ฎ๐น ๐ท๐ผ๐ฏ
rabbit inc. recently shipped a way to connect the R1 to OpenClaw, so the device can talk to an agent gateway instead of just the stock assistant:
https://www.rabbit.tech/blog/connect-your-r1-to-openclaw
Cool idea โ but two things got in my way:
1. The R1 only speaks the OpenClaw โclawdbot-gatewayโ protocol.
2. That path leans on provider API keys. I donโt live in API keys โ I live in Claude Code (Anthropic) and Codex (OpenAI) every day, signed in with my normal subscriptions (OAuth/login, e.g. OpenAI), not raw keys.
So instead of fighting it, I leaned into it.
I wrote a small WebSocket shim that impersonates an OpenClaw gateway. The R1 thinks itโs talking to OpenClaw โ but every push-to-talk turn actually gets routed to whatever agent is running on my server (an Apple Mac Mini M4 sitting on my desk):
R1 โ Wi-Fi โ shim on the Mac Mini โ local Claude Code / Codex CLI โ answer streamed back to the R1 screen.
Get Saurabh Sarkarโs stories inย yourย inbox
Join Medium for free to get updates fromย thisย writer.
The nice part: it reuses the CLIโs own auth, so it just works with my Claude and Codex logins. No API key juggling. The R1 becomes a dedicated hardware push-to-talk remote for the coding agents I already use.
Bonus gotcha for anyone trying the official path: current OpenClaw floors its gateway at protocol v4, but shipped R1 firmware only speaks v3 โ so a fresh gateway rejects the very device itโs meant to serve (โgateway unreachableโ). My shim negotiates down to v3, and I documented the downgrade fix for the official route too.
Open-sourced it (MIT), Claude Code + Codex backends included:
https://github.com/sarkarsaurabh27/rabbit-r1-hooks ๐ค
Repurposing hardware you already own > buying more hardware. Finally using my Rabbit R1 for something. ๐
#rabbitr1 #claudecode #codex #aitools #opensource #buildinpublic #hackathon







