Run a bounded repository task in a reproducible cloud environment.
Coding & development · Documented · Documentation checked 2026-10-04
Codex CLI, IDE extension, and desktop coding work; cloud environments need separate setup.
ChatGPT entitlement or supported API authentication, host capabilities, and workspace policy affect access.
An illustrative recipe. This website does not run the automation.
The goal and the right kind of system
Run a bounded repository task in a reproducible cloud environment.
A cloud agent fits remote development with known setup. A fixed host process should enforce budget and exit behavior for repeated runs.
Inputs and connected services
Inputs: authorized repo, prepared environment, scoped test credentials, and acceptance criteria.
Services: eligible Codex / Work cloud environment and authorized source provider; no production deployment authority.
The steps
An illustrative sequence to adapt and test before using real data or enabling ongoing work.
- Verify published environment setup and dependencies.
- Start a new scoped task and confirm branch / context.
- Inspect checks and generated changes.
- Run a review and hand off evidence for human merge decision.
Outputs and human checkpoints
Outputs: reviewable changes, setup / test logs, and blockers.
Human checkpoints: approve environment access and review actual diff / test outcomes.
Exceptions, permissions, and stopping conditions
Permission boundary: remote repo and approved test resources. Existing tasks can retain different environment state.
Stop on failed setup, missing services, unknown secrets, or unexplained changes. No merge or deploy follows automatically.
Adaptable prompt
Complete [scoped task] in the prepared cloud repo. Verify environment and branch, run actual checks, and produce a reviewable diff. Report blocked dependencies. Do not merge, deploy, or widen credentials.
Related reading
Cloud environments and repository access
Prepare a repeatable remote environment with dependencies, scoped secrets, and a known test command.
Reviews and repeatable noninteractive runs
Use a code review or noninteractive task as evidence for a human-controlled change.
Shell tools, MCP, and permissions
Distinguish a requested command from the filesystem, network, and external access actually allowed.
Designing an AI workflow
Make a useful task repeatable with clear inputs, steps, outputs, exceptions, and human checkpoints.
Permissions and meaningful approvals
Decide what a system may read, what it may change, and when someone needs to authorize a concrete action.
Sources and verification
Documentation checked 2026-10-04. Examples are illustrative unless described otherwise.
Cloud environments · OpenAI
Get started with Work · OpenAI
Codex CLI · OpenAI
Scheduled tasks · OpenAI
Permissions and approval policies · OpenAI
Plugins and supported surfaces · OpenAI