docs: CHANGELOG stage-358 — close v0.51.64 (stage-357), open Unreleased for 2-PR stage-358 (#2099 opt-in streaming text fade + #2165 pooled Codex quota status)

CHANGELOG.md

@@ -2,6 +2,14 @@
 
 ## [Unreleased]
 
+### Added
+
+- **PR #2099** by @dobby-d-elf — Adds an opt-in `Settings → Preferences → Fade text effect` toggle (off by default). When enabled, newly streamed output tokens are revealed through an adaptive playout buffer and animated with an opacity-only fade similar to ChatGPT and other frontier LLM apps. Implementation details: fade locked per stream to avoid mid-stream toggle rewind; reduced-motion users get non-animated text; live cursor hidden while fade is active; custom renderer on `streaming-markdown` parser wraps only newly-appended words; animated spans replace themselves with plain text on `animationend` (no long-lived wrapper buildup in long responses); unsafe streamed `href`/`src` values blocked in fade renderer `set_attr` path. Performance tuning: 200ms base fade duration scaling to 350ms for fast output, 16ms word stagger, 320ms done-drain wait cap, 160 wps visual cap, max 2-3 words/frame, brief pauses after sentence punctuation. Default-off means existing users see no change. 293-line regression test pinning the contract.
+
+- **PR #2165** by @starship-s — Pooled OpenAI Codex quota status surfaced in the Providers panel. Pre-fix, the Providers page presented Codex quota as if there were only one credential/account state, which was misleading when users authenticate through a credential pool with several usable credentials, temporarily exhausted credentials, failed probes, and different reset windows. Now the active provider quota card includes a credential-pool summary (available / exhausted / failed / checked counts), displays the best currently-available pool windows in the collapsed view as "Best of N", and exposes per-credential detail behind an expandable section. Exhausted credentials are intentionally NOT re-probed while their cooldown is active (matches credential-pool selection behavior, avoids generating failed quota calls from a status page). Manual refresh still means "probe now" but transient refresh failures preserve the last known-good snapshot. JWT decode (`_decode_jwt_claims_unverified`) is used only for token-shape classification (Codex OAuth JWT vs raw OpenAI API key), explicitly NOT for authorization — documented in the function docstring. Per-row plan labels only shown when verified account-limit data is available. Concurrent probing capped at `min(_CODEX_POOL_MAX_WORKERS=6, len(probe_items))` so page render time stays bounded on large pools. Transient `None` probe results are NOT cached (only known unavailable/exhausted states are cached); 32-test regression suite covering pool snapshot, concurrent probe, JWT detection, cache invalidation, transient-vs-known cache distinction, and i18n parity across all currently-supported locales. Scoped to OpenAI Codex (the only provider with the credential-pool/account-limit path needed to surface this accurately).
+
+## [v0.51.64] — 2026-05-14 — Release AN (stage-357 — 3-PR small batch — docker_init k8s whoami fallback + PWA manifest session routes (closes #2226) + aux title test coverage)
+
 ### Fixed
 
 - **PR #2270** by @Michaelyklam (closes #2226) — Firefox Android PWA installs from `/session/<id>` pages now resolve the Hermes manifest and icons instead of falling back to a generated letter icon. The dynamic `<base href>` script now runs before manifest/favicon links, `/session/manifest.json` and `/session/manifest.webmanifest` return the real manifest JSON, and session-prefixed manifest routes are now marked as public auth-skip routes. Adds 211 lines of regression coverage for the manifest responses and the session-prefixed 512px icon path.