Dive

Dive is an open-source Go framework that lets you build AI agents, agentic CLIs, and embed AI directly into backend services. Created by Deep Noodle, the team behind the Risor scripting language and Wonton CLI toolkit, it's designed for Go developers who want lightweight, customizable AI automation in production without heavy dependencies or vendor lock-in. You import Dive as a library, compose multi-step workflows, call language model APIs, and manage agent state all within your existing Go deployment patterns. Apache 2.0-licensed, it's available on GitHub and has over 130 stars. What sets Dive apart is its embeddability. Instead of a heavyweight platform, you get the primitives to fit your architecture. Build agentic CLIs that automate developer workflows, or embed agents into microservices to handle document processing, data pipelines, and internal operations. The Deep Noodle team actively maintains Dive alongside their broader work—custom agent builds, production readiness assessments, and AI security reviews—so it's backed by people shipping production AI systems daily. Dive is ideal for engineering teams already invested in Go who want a pragmatic path to agentic automation without pulling in a large dependency tree. You handle more integration yourself—there's no hosted control plane, so you're responsible for hosting and orchestration. The ecosystem is smaller than Python-based alternatives, but the trade-off is minimal dependencies and tight integration with your stack. For teams evaluating agent frameworks, Dive competes with LangChain and crewAI but takes a more focused, Go-native route. It's not a one-size-fits-all toolkit; it's a solid foundation for building your own agent logic with the Go idioms you already use.