It feels like their list of after actions is lacking a bit to me.
How about
1. The permissions change project is paused or rolled back until
2. All impacted database interactions (SQL queries) are evaluated for improper assumptions or better
3. Their design that depends on database metainfo and schema is replaced with ones that use specific tables and rows in tables instead of using the meta info as part of their application.
4. All hard coded limits are centralized in a single global module and referenced from their users and then back propagated to any separate generator processes that validate against the limit before pushing generated changes
How about 1. The permissions change project is paused or rolled back until 2. All impacted database interactions (SQL queries) are evaluated for improper assumptions or better 3. Their design that depends on database metainfo and schema is replaced with ones that use specific tables and rows in tables instead of using the meta info as part of their application. 4. All hard coded limits are centralized in a single global module and referenced from their users and then back propagated to any separate generator processes that validate against the limit before pushing generated changes