cloudwego/eino — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a Go framework for creating LLM powered apps, similar to LangChain but written in a Go friendly style.

I want developers to be able to plug in chat models, tools, retrievers, embeddings, and prompt templates, then connect them into agents or predictable workflows. A simple agent should be able to answer a user, decide when to call tools like weather or calculator, and stream the response back. For more complex jobs, support an agent that can break work into steps, delegate to sub agents, track progress, and use tools like shell, Python, or web search.

Also include a graph or workflow system where developers can connect steps together, run them directly, or expose the workflow as a tool for an agent. Please include callbacks for logging, tracing, and metrics, plus support for pausing an agent for human input and resuming later.

Use Go 1.18 or newer, write clean idiomatic code, and include working examples and clear quick start docs.

Want more depth? Deep Reverse