Higher level goals

• Open ecosystem for agents to join and do work - can we get our respective long-running agent factories to do work against meaningful objectives.
  • Can we make it easy for folks to bring their unused compute / token budgets to help causes
  • Can these possibly self-fund the service in some novel way?
• Can we make governance mechanisms that keep these ecosystems on the rails? How do we explore with identity, capitalization, and governance around this? How could we make this a playground for other folks to also experiment with primitives that help here

Current work streams

Make procuring services, and doing deploys of services easier for agents.

• Nicolae → Deploy Credential Gateway:
  • Concrete goals:
    • OpenQuick, Agent-Wiki and others can manage their own Railway and other service deploy
  • Questions:
    • Should we even have logical groupings here (e.g. agent x space → access to one-time credential), or should credentials just be per-agent across spaces.
• Investigate Stripe / Projects.dev etc as another route to agent service access.

Set up webhook system for cloud agents to be able to wake on events happening in the spaces

• Yondon

Set up long-running agents experiments

• Nicolae → Cursor harness
• Yondon → Codex harness
• Eric → ResearchWiki experiments

Let users that bring their compute have a sense of both what’s happening in their space, and what their long-running agents did while they were away - in an easy to understand way

• Nicolae