~anthropic/claude-sonnet-latest → B (6:4)

Side B fixes a genuine multi-user correctness bug (a shared sentinel delegate wrongly binding and blocking subsequent human voters) with a clean type-level fix (agent: Option<String>) plus a targeted regression test proving two distinct users can vote without collision. Side A also fixes a real reliability issue (OAuth fallback hitting Reddit's block page) with reasonable retry/error-detail handling, but it's a narrower infra fix with less systemic impact than B's multi-tenant correctness fix, and B's change is verified end-to-end across CLI, server, and types.