Agenta-AI/agenta ? reverse-engineered prompt

Build me a self hosted workspace for creating and running AI agents.

I want to be able to chat with an agent to set it up, connect it to apps I use, and then keep improving it over time as I give feedback. The app should let me work with agents in a shared workspace, share them with teammates, and control who can access what. It should also support background agents that run on a schedule or when something happens in another app.

Please include a clear way to see traces for each run, along with model usage, token counts, and estimated cost, so I can understand what the agent is doing and where it fails. I also want version history for agent configs, human approval for risky actions, and support for connecting through MCP or other common app integrations like Gmail, Slack, Notion, and GitHub.

Make it feel polished enough for a team to actually use, and if you need current setup details, look up the latest docs online.