Deprecation context for AI coding agents, over MCP
An AI coding agent will write whatever the codebase makes look normal. If a deprecated API is still present in a thousand places, that is what the agent learns from — and it will add the thousand-and-first call.
Agents lack organizational context about what is deprecated, why, and what to use instead. Obsolot supplies it.
An MCP server over your deprecation registry
Obsolot exposes its notices and code-reference data over the Model Context Protocol. An agent connected to it can query what's deprecated, retrieve the migration guidance attached to a notice, classify code references, and create migration suggestions — in real time, while it works.
The same data the review queue runs on is the data the agent sees. There is no separate export to keep current.
The agent proposes, a human approves
Suggestions created by an external agent land in the same migration suggestion review queue as the ones from Obsolot's own embedded agent. Nothing is applied because an agent said so; every proposal is approved or rejected by a person.
That boundary is deliberate. Giving agents context is useful precisely because the output still gets reviewed.
Why this matters more as agents write more code
The rise of AI-assisted coding introduces a new participant in the migration workflow — one that is fast, tireless, and completely unaware of your end-of-life dates unless you tell it. A deprecation registry that agents can query is how you keep the acceleration pointed forwards.
See how notices are published on the deprecation management page, or compare plans.