4 How the Migration Skill Validates Results

Producing a migrated project is only half the job. Code migration itself can be lengthy (potentially hours), so the process was designed to be resumable in case something goes wrong with an AI environment or the process needs to be stopped partway through. Once migration is complete, the resulting application must still be proven to work on real hardware. This is handled by a separate test migration skill, invoked after code migration, through a three-part process:

• Orchestrator: Determines what test equipment is available and leads the user through any required setup (wiring, jumpers).
• Planner: Reads the migration report and writes a prioritized test plan.
• Executor: Runs each test, applies fixes on failures, rebuilds, and retries (up to a bounded number of attempts) before flagging the issue for a human.

Results are logged as they are received, giving visibility into test progress. The approach is also adaptable to what is on the bench and can drive an existing test framework and lab equipment such as a logic analyzer or signal generator, as well as on-target debug and serial communication. The migration skill produces a test suite that can be reused separately from the AI environment.