This Work conversation already has tools for running code, so a short public hypothesis test does not require another API agent. An external Agents API backend becomes useful for separate asynchronous research jobs with their own sessions and programmatic lifecycle.
Public experiments during a Wiki discussion
The present Work conversation exposes command-execution tools. A requested short public experiment can therefore run here, with the code, assumptions, results and limitations recorded in the research.
Work and Codex document sandboxed command execution and cloud development environments. Available access still depends on the environment and permissions.
A separate Agents API integration is useful when an experiment needs an independent asynchronous job, a session managed by application code, or reusable execution setup. It adds a backend and API billing; it is not necessary simply to obtain code execution in this conversation.