VibeKit is an open-source SDK to run coding agents like OpenAI Codex and Claude in secure sandboxes. Let agents write code, install packages, or open PRs safely — with streaming, async tasks, and telemetry built-in. MIT licensed, TypeScript, zero lock-in.
Hi Product Hunt 👋
We’re Alan and Ismail, the makers of VibeKit (from Superagent, YC W24).
We built VibeKit because we were using AI coding agents (like Codex and Claude) heavily, but kept running into pain when trying to run them safely — especially in apps, CI flows, or GitHub bots.
VibeKit is a drop-in SDK that runs coding agents in secure sandboxes (like E2B). It’s open-source, TypeScript, model-agnostic, and designed to help you ship real features — not just local experiments.
We’d love to hear your thoughts, try it out and let us know what use cases it unlocks for you.
VibeKit strikes the right balance between flexibility and safety — letting coding agents work freely in secure environments. Love the open-source ethos, async support, and dev-friendly setup. A solid foundation for agent-powered tooling!
Scribe
No human-in-the-loop? In general, in have found it difficult to get results at the right productivity levels without human in the loop
Super RAG
@manu_goel2 We are planning to add human-in-the-loop functionality in the future.
VibeKit strikes the right balance between flexibility and safety — letting coding agents work freely in secure environments. Love the open-source ethos, async support, and dev-friendly setup. A solid foundation for agent-powered tooling!