Run an AI agent locally on your own computer using your own API keys instead of a hosted service.
Break a large task into smaller pieces handled by coordinated worker agents.
Connect MCP servers and install skills so the agent can extend its own abilities mid task.
Keep a persistent conversation about a local project that survives long sessions without losing the original goal.
| reina-agent/reina | 0xovo/litedoc | 7etsuo/anvil | |
|---|---|---|---|
| Stars | 26 | 26 | 26 |
| Language | — | HTML | TypeScript |
| Setup difficulty | easy | easy | moderate |
| Complexity | 3/5 | 1/5 | 3/5 |
| Audience | developer | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires bringing your own AI model provider and API key.
Reina is a desktop app that acts as an AI agent for your local projects. Instead of chatting in a browser tab, you install it on Windows or macOS and it holds an ongoing conversation about your files and tasks, helping organize information and push multi step work forward. You provide your own AI model and API key in the app's settings, and those keys are stored only on your machine, never sent through any company server owned by the project. One of its main ideas is working as a team. When a single agent is not enough, a coordinator agent breaks a goal into smaller connected tasks and directs a set of worker agents through the steps until each part is finished, rather than stopping early and calling the job done. The app also supports something called MCP servers and Skills, which are add ons you can connect or install from built in marketplaces, and the agent itself can look for and add new skills while working on a task. Before any action that could change files or systems, Reina asks for your approval, and each step it takes is visible so you can pause or resume the process. The app is organized into separate windows for chatting, viewing your workspace, and adjusting settings, and it updates itself automatically once installed. To try it, you download an installer for your operating system from the project's releases page, pick a workspace, enter your model and API key, then give it a task. Building it from source requires Node.js and involves cloning the repository and running install and build commands. Reina is released under the AGPL-3.0 license, meaning you can use, change, and share it freely, but any modified version, including one offered as an online service, must also share its full source code. The author offers a separate commercial license for closed source use.
A desktop AI agent app that manages local projects, coordinates teams of sub-agents, and connects to your own chosen AI models.
Free to use, modify, and share, but any modified version, including one run as an online service, must release its full source code under the same license.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
double-check against the repo, no cap.