Repolode
Submit
Ironclad

Ironclad/rivet

Visual IDE for building AI agents and prompt chains as graphs, with a TypeScript library to embed them in your own apps.

View on GitHubWebsite
Stars
4,719
Forks
393
Open issues
109
Last commit
2 months ago

What is it?

Rivet is a desktop application for designing AI agents and prompt chains as visual node graphs. Users wire together prompts, LLM calls, embedding lookups and other steps in an editor, then run the resulting graph to test its behavior.

The project also ships Rivet Core, a TypeScript library (published as @ironclad/rivet-core) that runs graphs created in the Rivet editor. Application code can call into a graph, and graphs can call back into application code, so the same logic built visually can be embedded in a product.

The application supports OpenAI GPT-3.5 and GPT-4, several Anthropic Claude models, and AssemblyAI's LeMUR and speech-to-text services. For embeddings and vector search it integrates with OpenAI Embeddings and Pinecone. Prebuilt binaries are available for macOS, Linux and Windows.

Key features

  • 01Visual node-graph editor for composing prompts and LLM calls
  • 02TypeScript library for running graphs inside your own applications
  • 03Supports OpenAI GPT-3.5/GPT-4 and Anthropic Claude 2 and Claude 3 models
  • 04Embedding and vector search integrations with OpenAI Embeddings and Pinecone
  • 05Speech-to-text and LeMUR voice analysis via AssemblyAI
  • 06Desktop builds for macOS, Linux and Windows

Use cases

  • Prototyping multi-step prompt chains before writing code
  • Building AI agents whose logic non-engineers can inspect visually
  • Embedding visually designed LLM workflows into a TypeScript or Node.js product
  • Combining voice transcription with LLM analysis pipelines

Quick start

  1. 1

    Download the prebuilt app for your platform from the latest releases

  2. 2

    To build and run from source, follow CONTRIBUTING.md

Things to know

  • Built-in LLM and embedding support covers only the providers listed in the README.

In collections

Claude Code tools, skills & pluginsOpen-source AI agent frameworks & apps

Topics

#llm#ai-agents#prompt-engineering#visual-programming#typescript#openai#anthropic#vector-database#ai

FAQ

What is rivet?
Visual IDE for building AI agents and prompt chains as graphs, with a TypeScript library to embed them in your own apps. Rivet is a desktop application for designing AI agents and prompt chains as visual node graphs. Users wire together prompts, LLM calls, embedding lookups and other steps in an editor, then run the resulting graph to test its behavior.
Is rivet free to use? What license does it have?
rivet is open source under the MIT license. Check the license terms before using it in commercial products.
Is rivet still maintained?
Updated within 6 months. The last commit to rivet was on Aug 26, 2026.
What can I use rivet for?
Prototyping multi-step prompt chains before writing code Building AI agents whose logic non-engineers can inspect visually Embedding visually designed LLM workflows into a TypeScript or Node.js product Combining voice transcription with LLM analysis pipelines
How do I install rivet?
Download the prebuilt app for your platform from the latest releases To build and run from source, follow CONTRIBUTING.md
Where can I get rivet?
The source code and installation instructions are on GitHub: https://github.com/Ironclad/rivet

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects