Pipelining a Safe Mass-Fix When the Audit Finds Six More Instances

In the case study, the Phase 5 generalization workflow finds exactly one more instance of the missing-tiebreaker pagination bug (the audit-log export pipeline), which is fixed by hand the same way as the original export-engine bug.

Now suppose the same workflow, run against a larger version of this monorepo, finds six more instances across six different files, each with a slightly different cursor implementation (different column names, different ORMs in two cases, one raw SQL query).

1. Would you still fix all six by hand, one at a time, the way the case study fixed the single audit-log instance? Explain why or why not, referencing what changes about the task at six instances versus one.
2. Design a pipeline (using a dynamic workflow) that fixes multiple instances safely, given that the fix isn't a pure mechanical find-and-replace — each site needs a tiebreaker column identified, a composite comparison written in whatever style that file already uses, and a covering index added.
3. What verification step is non-negotiable in your pipeline, and why is it more important here than it was for the single hand-fixed instance in the case study?