At the recent GitHub Universe 2025 developer conference, Arm unveiled the Cloud migration assistant custom agent, a tool designed to help developers automate, optimize, and accelerate the migration of their x86 cloud workflows to Arm infrastructure.
According to Arm, their new migration assistant leverages AI to “understand the why behind each migration step”. The assistant is integrated within GitHub Copilot and relies on the Arm MCP server to interface directly with the actual engineering tools, performance libraries, and migration utilities.
Your goal is to migrate a codebase from x86 to Arm. Use the mcp server tools to help you with this. Check for x86-specific dependencies (build flags, intrinsics, libraries, etc) and change them to ARM architecture equivalents, ensuring compatibility and optimizing performance. Look at Dockerfiles, versionfiles, and other dependencies, ensure compatibility, and optimize performance.
The assistant can scan a GitHub repository to detect architecture assumptions, portability issues, container base images, and dependency incompatibilities. Based on this analysis, it provides recommendations to optimize your code for the Arm architecture.
More specifically, the assistant examines all Dockerfiles and uses check_image and/or skopeo to verify ARM compatibility. If a base image is not compatible, it can select an alternative. It also inspects all installed packages, both those installed by Docker and those listed in requirements.txt, using the learning_path_server tool, replacing any incompatible packages with ARM-compatible versions. Finally, the migrate_ease_scan tool is used to analyze the codebase, identify required changes, and apply them.
After making the necessary modifications, the assistant can rebuild the entire project for the Arm architecture and run any available benchmarking or integration tests.
The Arm MCP server, which does all the heavy lifting in the workflow described above, is currently available in beta. It specifically supports AWS Graviton, Google Axion, and Microsoft Azure Cobalt 100-based virtual machines. Among the tools provided by the server are Sysreport, which analyzes a system to ensure it is configured for performance analysis; Machine Code Analysis, which evaluates performance of compiled code; Arm Knowledgebase Search, and several others.
Developers can also access the MCP server to build custom workflows using the provided tools.
Although Arm has just announced its integration with GitHub Copilot, the Arm MCP server also works seamlessly with any other client supporting the MCP protocol, including AmazonQ, Claude, Cursor, WARP.dev and others. To maximize the benefits from the Arm MCP server when using a client other than the Arm Copilot migration assistant, you can leverage the system prompt that the assistant employs.
The Cloud Migration Assistant is available in GitHub’s awesome-copilot repository with early access registration.
