# Running Claude Code and Codex as a team in a shared room: what worked (built with SharedNet) > Source: > Published: 2026-10-04 20:57:57+00:00 I recently took part in Trial Zero, a hackathon run on SharedNet ([https://www.sharednet.ai](https://www.sharednet.ai)) by Systemind. The idea: put coding agents from different vendors into one persistent room where they talk, hand off tasks and even pay each other, then let them compete with no humans at the keyboard. What I built: Scout, an MCP tool that checks whether an agent product is actually callable before you pay for it. It scores a product /100, lists the fixes that block a first call, makes one real call to a safe read-only tool on its live endpoint, and states what it did NOT verify. Repo: [GitHub - yeziR4/scout · GitHub](https://github.com/yeziR4/scout) · How the agents collaborated in SharedNet: In the arena round, agents from other teams bought Scout’s services with credits, and some found real bugs live. Our agent acknowledged each one in the room and shipped a fix within minutes. We placed 3rd. If you’re experimenting with multi-agent setups, SharedNet is worth a look as a neutral “room” between agents from different vendors. Happy to answer questions about the setup. #TrialZeroHackathon