Noumena Code

Noumena launched NCode, a platform that uses cloud agents to write code within scoped workspaces, managed via a terminal UI and stack-based reviews. The system aims to coordinate multiple AI agents working on the same repositories, keeping human oversight in the review and landing process.

We tried to use agents to write code. The hard part was managing them. cloud workspacencode tui / agent task Scoped workspace Cloud agents TUI control @ qlpvwonx jj draft │ tune ranking signals │ o krunqpzt sl commit │ add issue fixtures │ o pnyxmtkl sl commit ╰─ expose repo search o main production JJ draft · SL commit stack · GitHub-compatible review NCode is deeply integrated with the terminal UI and the agent runtime. Agents can run in the cloud against native scoped workspaces, while humans keep review and landing decisions in one place. We went back to how large engineering organizations coordinate many workers changing the same repos. Issues create scoped workspaces. Agents produce drafts. Review happens as stacks. The platform remains GitHub-compatible, but the native workflow is built around agent work, cloud execution, and safe review. TUI control start, inspect, and steer agent work from the terminal Cloud agents run agents remotely without losing repository context Scoped workspaces keep file access and work boundaries explicit Stack reviews review related changes together before landing Compatibility keep familiar repository, issue, and review surfaces