Next#

This page lists what Booster does not do yet, and what it decided not to build. It is the only place future work lives; the other pages describe what ships today.

Open#

• The gate needs a real app. The contract tests cover the bundled example apps. A full run against a real Wayne State application with the gate on has not been proven in CI yet; the proof site carries those runs.

Decided not to build#

• An MCP server for Booster. App owners drive upgrades from their own machines and CI; an MCP endpoint would add a surface to secure without a caller that needs it.
• Automatic merge requests. Booster edits a working copy and stops. Opening MRs on app repositories would make a tool that fixes code into one that ships it, and the person reading the diff is the point of the review threshold.

Not built yet#

• Custom rules. Apps sometimes need a rewrite Rector does not ship. Booster has no way to register one yet; until it does, such rewrites stay manual.
• Pilot numbers. No pilot cohort has run yet, so there are no counts of apps upgraded or hops applied to publish.
• A cas-to-sesame run against a real app. The hop exists (opt in with --sesame) and renames the classes, swaps the Composer requirement, writes config/sesame.php from the old config and rewrites the model trait. Duo two-factor stays manual work, and the hop has not been run against a real app yet.