On this page · jump to a step
Connection capabilities and usage rules are separate
| Relayard provides | Your account and local environment determine |
|---|---|
| A connection to an authorized local Xcode project. | Chat Pro availability, remaining usage and tool support. |
| Delivery of actual tool operations and their results. | Completion, the number of requests and whether builds pass. |
Relayard does not reset, expand or bypass model limits. It uses Chat Pro usage already available to your account. There is no fixed monthly message promise or guarantee of a release in one request.
When is this a poor fit?
If Chat Pro is also at its limit, the model lacks the required tools, account policies prevent connection or the project cannot run locally, this workflow does not remove those limits. Wait for a reset or use another eligible option with the capabilities you need.
After reaching a Codex limit, you can use Relayard to continue in the same local Xcode project with Chat Pro, provided you have usage remaining and the required tool support.Chat Pro has a separate allowance from Work / Codex. You still need a project handoff and a working connection.
Hand off the actual project state
Imagine favorites are half implemented in an iOS app, with filtering, an empty state and a build check left to finish. Save the current files and a Git checkpoint, then record completed work, remaining tasks, the latest error and acceptance criteria. Relayard lets ChatGPT inspect the same project; it does not import the Codex conversation.
Define the full scope of this iteration
One request can organize planning, changes and verification. Specify features, boundaries and acceptance together instead of just asking the model to continue. A scoped iteration of an existing app can follow this workflow; missing permissions, complex errors or limited context may still require follow-up requests.
Add favorites and filtering to the current app list. Inspect the project and its existing list, storage and UI conventions. Scope: toggle a favorite, filter favorites, and show an empty state. Preserve navigation and data. Make a short plan, then work through the related changes. Use authorized builds, tests or previews to check results. Acceptance: favorites persist using the existing storage approach; filtering and empty states work. Confirm the binding, permissions and actual tools first. Preserve original errors; never invent a pass. Report changes, verified results and unfinished work. Do not automatically commit or publish.
When is this useful?
- You have an existing Xcode project that opens on your Mac.
- You have Chat Pro usage remaining and account and model support for the required tools.
- You can define the goal and review changes and build results.
What needs to be ready first?
Get an Relayard app build, authorize the project locally and complete OAuth in the client. Verify the bound project with a read-only request, then enable only the edits and execution needed for the task. Files and builds stay on your Mac; requested content passes through the hosted relay and AI client.
Does one message equal a full release?
This is not a fixed billing unit or completion guarantee. A complete task can contain many steps, but message accounting, reset periods and the amount of work completed depend on your account and project state.
Official OpenAI model and usage guide
Documentation updated: 2026-10-08