Rivet

Rivet provides a visual, node‑based graph editor for constructing and managing prompt workflows that drive AI agents built on large language models. Users assemble chains of prompts and model calls as interconnected nodes, then execute those graphs directly within their applications. The environment includes a remote debugger that displays real‑time execution details, allowing developers to inspect the flow of data and diagnose issues while the system runs.

The tool is designed for teams that need to collaborate on complex LLM prompt graphs. Graphs are stored as plain YAML files, which can be version‑controlled in standard repositories and reviewed with existing code‑review processes. This format enables straightforward sharing, review, and iteration across multiple contributors.

Rivet is distributed as an open‑source project targeting macOS and is positioned for stable, production‑grade use. It aims to simplify the creation, debugging, and deployment of AI‑driven workflows without requiring extensive manual coding of prompt logic.