B fixes a real security/design flaw (feed cutoff keyed by unauthenticated actor string) with an auth-bound delegate model, adds thorough integration tests covering rejection paths, and updates CLI/docs consistently. A is a substantial refactor adding children-fetch support and SSE-as-JS-eval, which is useful but more speculative/UI-plumbing with less clear-cut correctness payoff and more churn (rewriting SSE protocol, JS handling) than B's targeted security fix.
Commit A delivers substantial product capability: separate self/children Reddit fetch kinds, children listing import with tree linking, OAuth token vs API base split, SSE unified to eval JS morphs, and ranking_panel by connected components—backed by fixtures and tests. Commit B is a focused, correct feed redesign (delegate scoping, auth/binding checks, principal-wide catch-up) with solid tests, but it is a narrower API/CLI behavioral fix versus A’s multi-layer feature and protocol work.
Side A delivers a substantial functional expansion by introducing separate fetch modes for entities versus children, wiring the new fetch kind through the UI, SSE pipeline, Reddit worker, state management, and adding child-import parsing/linking plus updated ranking behavior for imported children. Side B makes an important API/security improvement by scoping feed retrieval to authenticated delegates and adding authorization checks and tests, but it is narrower in scope than A's end-to-end feature and architectural changes.