I’ve been working on Opperator, an open source framework for building and running general purpose AI agents locally from the terminal. It’s similar in spirit to Claude Code, but focused on automation rather than just coding.

You can create agents that organize files, generate or transform content, monitor APIs, or automate personal workflows. Each agent runs as its own local process with its own environment and can use any model you want, including local LLMs.

A background daemon handles everything: it manages agent lifecycles, logging, persistence, and secrets. On top of that, there’s a terminal interface for interacting with agents and a lightweight Python SDK for defining their logic.

For example, you can say “create an agent that looks at my screenshots folder and renames files based on what’s in the image.” The built-in Builder agent scaffolds the code, installs dependencies, and lets you iterate live without restarting. Of course, you can also write or refine agents manually with tools like Cursor, Codex, or any other code assistant.

GitHub: https://github.com/opper-ai/opperator Docs: https://docs.opper.ai/opperator

I’m interested in whether this idea resonates with others who like working locally and building personal automation systems. Feedback, use cases, or architectural critiques are all welcome.