Hi Product Hunt I have been building Rebuno, an open-source execution runtime for AI agents. Rebuno brings durable execution state, guardrails, and an audit trail into a shared runtime for agents built with different frameworks. Agents run as HTTP services, with Rebuno coordinating their execution. Each connected tool or model call becomes a recorded step. Policies determine whether it can proceed, is denied, or needs approval. When a run resumes after an interruption, completed steps return their recorded outcomes. If a tool's outcome is unknown, Rebuno retries it only when the tool is configured as safe to repeat. Otherwise, it flags the uncertainty for the agent to handle. There are Python and TypeScript SDKs, and the repo includes examples using LangChain, CrewAI, Pydantic AI, Vercel AI SDK, and Mastra. Rebuno is free to self-host and released under the MIT licensed. The blog post explains the design with diagrams: https://rebuno.io/blog/introduci... I'd be interested in hearing how others are running agents in production.

Rebuno

Durable execution and guardrails for AI agents

Durable execution and guardrails for AI agents