Plan a software migration with AI
Identify consumers of an old interface, check whether its replacement preserves the behaviours they need and define the evidence required before retirement.
To plan a software migration with AI, investigate who still depends on the old behaviour. A client may rely on the ordering of a response even if the original interface never promised it. The deprecation-and-migration skill treats retirement as a change for real consumers, with a replacement path and measurable exit conditions.
A deadline can motivate the move but cannot tell you whether it is complete. The plan can distinguish clients already migrated, occasional jobs that need a longer observation window and cases where running both paths temporarily makes the change easier to reverse.
What it does
- Inventory known consumers and check for undocumented dependencies.
- Compare the old and new behaviour from a caller’s perspective.
- Prepare communication, migration instructions and a timetable.
- Stage a cutover and monitor evidence of remaining use.
- Define removal criteria, rollback options and post-migration cleanup.
Install and use
- Install the skill and supply the existing interface, proposed replacement, consumer information and authorised usage evidence. Live configuration or deployment changes require separate access to the project.
- Ask, “Plan retirement of our old orders endpoint. Show the behaviours callers may depend on and the evidence we need before removal.”
- Review the plan with consumer owners, observe the agreed period and check the final retirement criteria.
FAQ
Source
Made by Addy Osmani. Original skill: deprecation-and-migration, published on GitHub under the MIT licence.
View on GitHub