> ## Documentation Index > Fetch the complete documentation index at: https://docs.modelcode.ai/llms.txt > Use this file to discover all available pages before exploring further. # Use Cases > Run your Morph projects, review milestones, and re-run reviews or rebase from Claude Code > [!NOTE] > > **Now in beta.** The Claude Code integration is rolling out to a select group of customers. Want early access? Reach out to [support@modelcode.ai](mailto:support@modelcode.ai) and we'll get you set up. Once the plugin is [set up](https://docs.modelcode.ai/claude-code/setup.md), talk to Claude Code as you would to a teammate who has access to the Morph Platform. Every run you start and every change you make shows in the Morph UI, so you and your team can switch between the two at any time. *** ## Run a project ### See your projects and switch between them > "What are my Morph projects?" > > "Switch to the payments project." Claude Code lists the projects in your organization, with a link to each in the Morph UI. Every other request applies to the project you're working on, until you switch. ### Ask where things stand > "What's the status of the project?" > > "Why can't milestone 3 start?" Claude Code reads the current state from the Morph Platform, never from memory: each milestone's state, its prerequisites, its runs with their task progress and errors, its pull requests, and links into the Morph UI. If the project isn't ready to run milestones yet, it says where the project is, for example a [Project Spec](https://docs.modelcode.ai/setup/reviewing-project-spec.md) waiting for your approval or a [Roadmap](https://docs.modelcode.ai/migration/roadmap.md) not generated yet, with a link to act on it in the Morph UI. It also tells you when the project's build environment is still being prepared or failed to build. ### Start a milestone > "Start milestone 1." > > "Start milestones 2 and 3." Claude Code first checks that the milestone can start: its prerequisite milestones are merged, it isn't already running, and the build environment is ready. If something blocks it, Claude Code tells you what and doesn't start it. On a billed plan, Claude Code shows you the milestone's price in credits and starts only after you approve it. Milestones that don't depend on each other can run at the same time. ### Follow a run > "How is milestone 1 doing?" > > "Tell me when milestone 1 is done." A milestone run goes through three steps: the agent run, its automated review, then the pull request report. Claude Code tells you which step a run is at, its task progress, and the agent's latest progress report. Once all three steps succeed, the milestone is ready for code review, and Claude Code gives you its pull requests and a link to its **Code Review**. > [!TIP] > > To have Claude Code keep checking without you asking each time, use `/loop`. For example: `/loop every 5m, check milestone 1, and tell me when it finishes`. If the agent needs your help to continue, for example a secret it can't find, the run pauses and waits for you. Claude Code lists the blocker issues it waits on, with a link to the help chat in the Morph UI where you work through them. The run resumes by itself once they are resolved. ### Retry a failed step > "Milestone 2 failed. Why?" > > "Retry it." Claude Code tells you which step failed and why, then re-runs that step, keeping the work already done, as the roadmap's **Retry** button does. Retrying the agent run is priced like a start: on a billed plan, Claude Code shows you the price first. ### See where the final code delivery stands > "Where does the final code delivery stand?" > > "Give me the final pull requests." Once every milestone is merged, the Morph Platform opens the [Final Code Delivery](https://docs.modelcode.ai/migration/pull-requests.md): one pull request per repository, from the project's feature branch into the repository's default branch. Claude Code tells you where it stands, as the roadmap's Final Code Delivery card does: * **Not yet:** some milestones aren't merged. * **Held back:** open ad hoc, validation or sync milestones may add more work. Finish or delete them in the Morph UI first. If the delivery pull requests are already merged, the Morph Platform doesn't open a second delivery: bring that work into your default branch with your own pull request on your git provider. * **Being prepared:** the Morph Platform is opening the pull requests. * **Ready:** the pull requests are open, and Claude Code gives you their links and a link to review them in the Morph UI. * **Delivered:** the pull requests are merged. > [!NOTE] > > Claude Code doesn't merge the Final Code Delivery pull requests, on purpose: they bring the whole project into your default branch, so you review and merge them on your git provider, as in the Morph UI. *** ## Review a milestone Go through a milestone's code review from Claude Code: see where it stands, read its review issues and discuss them, then re-run the review or rebase the milestone. ### Explore the review > "In milestone 1's review, what are the open issues? Which ones are the most important?" > > "Explain issue 4. Is it a real bug?" Claude Code summarizes the milestone's review: the pull requests, the review issues with their status, and a link to the milestone's **Code Review** in the Morph UI. You can then go through the issues one by one and ask about them. ### Fix issues and rebase > "Re-run milestone 1's review." > > "Rebase milestone 2 with full validation." Claude Code starts these runs in the Morph Platform, which pushes any fixes to the milestone branch. Your local checkout doesn't change. A review re-run can fix the open issues that need a code change. For a single issue, use the review chat in the milestone's **Code Review**. A rebase uses **Basic validation** unless you ask for another level; see [Choosing how much a rebase validates](https://docs.modelcode.ai/migration/milestones.md#choosing-how-much-a-rebase-validates). If another run is working on the milestone's branch, Claude Code tells you it [can't start yet](https://docs.modelcode.ai/support/troubleshooting.md#cant-start-that-yet). *** ## What Claude Code can't do yet These still happen in the Morph UI: * Pausing or steering a running agent. * Generating the Roadmap, and approving the Project Spec. * Editing a milestone's plan. * Re-running validation or lifecycle commands on their own, outside a review re-run. * Reading or editing the project's wiki and rules. When you ask for one of them, Claude Code points you to the Morph UI. > [!NOTE] > > We're adding more capabilities to the Claude Code integration all the time. Stay tuned. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.