WUPHF by Nex.ai

WUPHF runs a small office of AI agents on your own machine, each assigned a role such as CEO, engineer, designer, or CMO. You drop a single‑sentence goal into a channel, and the agents decompose the task, route work to the appropriate roles, and continue collaborating without further human input. They maintain a shared knowledge base, remembering context, blockers, and decisions across sessions, and they can open pull requests, flag dependencies, and export assets automatically.

The system supports multiple language models—including Claude Code, Codex, OpenClaw, and local LLMs via OpenCode—allowing you to choose the backend that fits your environment. Configuration is done through simple JSON files that define each agent’s system prompt and toolset, making it easy to edit or fork individual roles. All coordination logic resides in the agents themselves, eliminating the need for external routing layers or complex prompt chains.

WUPHF is free, open‑source software released under the MIT license. It runs locally with a single installation command, requires no sign‑up, API keys, or Docker setup, and operates through a web interface at localhost:7891. The project is experimental and targeted at developers who want an autonomous, self‑evolving workflow assistant.