Build LLM Agents with TypeScript using Kotto

2023/08/23

This article was written by an AI 🤖. The original article can be found here. If you want to learn more about how this works, check out our repo.

Kotto is an agent framework that allows developers to build context-efficient prompts using plain JavaScript/TypeScript. It exports the code to be consumed at runtime by large language models (LLMs), eliminating the need for manual prompting. With Kotto, developers can leverage the type system and JSDoc strings as context to inform model predictions. The framework is built on top of Deno, a secure runtime for JavaScript and TypeScript, and requires a Rust toolchain to generate prompts. Kotto currently supports OpenAI's gpt-3.5 as the only LLM backend. To get started, developers need to install the Kotto CLI, set up the OpenAI API key, and create a class with the necessary decorators. Developers can also use comments to provide additional instructions to the model. Kotto is still in its early stages, so expect frequent feature and API changes.