Chloe

Chloe is an open-source AI agent framework for TypeScript developers who want their own code, not an LLM, to stay in charge of each step. Instead of handing a whole workflow to a model, you decide step by step which parts call an LLM and which run as plain code, which suits scheduled jobs, chat-based assistants, and automation where predictability matters. Triggers are built in: agents can fire on a cron schedule, respond on Telegram, Slack, WhatsApp, or email, or be invoked over an HTTP API. It ships with tools for Gmail, Google Calendar, Drive, web reading, and file operations, plus built-in step tracking that records cost per execution and supports spending and turn limits on AI calls, so a runaway loop has a ceiling. The whole thing self-hosts on a single SQLite database, which keeps operations simple for small teams. For a BestAIFor shortlist of the best open-source AI for building agents in code, Chloe's niche is clear: a self-hosted, MIT-licensed TypeScript framework for developers who prefer explicit control and full visibility. It is free to use; you pay only for the third-party LLM API calls your agents make, and it needs Node 22.18 or newer.

Building self-hosted TypeScript AI agents for scheduled jobs, chat assistants, and HTTP automations where code controls which steps use an LLM.