lemma-work/lemma-platform ? reverse-engineered prompt

Build me a local first open source workspace where people and AI agents can work together on recurring jobs. I want the full platform, a web app with an optional desktop app feel, a backend, and a CLI so the same coding agent can help create, run, verify, export, and update each workspace.

The main idea should be that every job gets its own app inside a pod. A pod should hold shared data, files, agents, workflows, permissions, and places where a person can review or approve what the agents want to do. It should be easy to start from a starter pod and then turn it into things like an expense tracker that works through Telegram, a research room, or a support desk. If possible, support chat surfaces like Telegram, Slack, Teams, and WhatsApp too.

It should run locally on my laptop with a simple install path, but also be able to run on a server or cloud later. Let me use my own model provider or existing coding agent login. Include clear docs and sane defaults, and look up current docs online if you need to.