I'm the founder of Tofu, a code deployment platform built for agents. Paste our setup prompt to any agent you're already using and Tofu enables your agent to take your project from code to production. Current features include frontend & backend hosting (created & managed by Tofu), Postgres database (created & managed by Tofu), domain management (one-click setup for Cloudflare managed domains, manual DNS record for others), authentication, Google login, and Analytics.

We built Tofu out of our personal frustrations: we're engineers who know how to code, we've deployed live products and understand the process, and our AI agents have already finished writing the code that works locally, yet taking a product fully online still took 2-3 hours: - Setup hosting, backend, database - Copy-pasting a lot of apikeys/secrets from here to there - Setting up correct redirect urls from places to places - Managing DNS records, domains - Setting up a lot services (email, 3rd parth OAuth, etc) - Payments

Tofu works with any MCP-compatible agents, and here's a list of tested agents (10+ different projects end-to-end deployment via Tofu): local coding agents (Claude Code, Codex/ChatGPT, Kimi Code, DSH, Cursor), Instinct, Muse, Kimi Code cloud.

None of that is hard, but each of these services have their own providers and it involves managing 5-10 different accounts. Most cloud providers have their own CLI, but it only works within their own platform and the process still involves a lot of manual interventions. Agents can call any API and use CLIs, but cannot open a Google Cloud project or put a card on file with 5 vendors.

We think deployment is not a compute problem, but an accounts problem. Our endgoal is to utilize SSO across services, and provide agents the ability to deploy a live product end-to-end. We don't want to compete with any cloud providers, but to make agents actually capable of shipping "live" products. We're rolling out features to support buy domains directly from Tofu (we're an authorized reseller of Name.com), and provide more customizability to tech stacks.

We're still very early and have just launched our product, please send us any feedbacks, comments, and we promise we will respond to every single one of them.