An open source platform to run many agents at a time. Control plane built on CloudFlare, with your agents running on VMs in your cloud provider of choice.
Been a while since I ran out of tokens. Restarting all with different agent. I wonder if I should build some tooling in @simpleagentman to detect this kind of thing and automatically reroute to an agent that has usage left.
Almost felt like a launch week at Cloudflare...
- Partnered with @xai to bring Grok to Cloudflare
- OpenAI announces Sites built with Cloudflare
- Acquisition of @voidzerodev & Vite team
- AI Gateway spend limits released
- Now support self-managed OAuth clients
🥵
We're building an open source meta harness sort of deal. But now I'm wondering if it makes sense to build a framework. The system itself can only do so much to guide your claude, codex, etc. instances...
@RonaldRateLimit@suni_code This does scare me. Though listening to the recent episode of @Pragmatic_Eng with @thdxr relieved me a bit: it seems like inference is a profitable business. So I assume a lot of the cost of operating the frontier labs is in training. As long as inference is cheap, I'm happy.
@nezbuilds An open source platform to run loads of AI agents in parallel in your own cloud accounts!
It runs on CloudFlare and lets you run Claude Code, Codex, OpenCode and other agents in VMs on different cloud providers :)
@lightpanda_io In yolo mode I just tell Claude to "use lightpanda to <some task in a browser>" and it seems very happy to use the markdown mode 😄