<!-- Release notes generated using configuration in .github/release.yml at release/v0.21.7-nightly.20260808.4ec0371e6 -->
## What's Changed
* fix(ci): surface blocked autofix takeover admission by @qqqys in https://github.com/QwenLM/qwen-code/pull/8410
* docs: document serve sub-session concurrency settings by @DragonnZhang in https://github.com/QwenLM/qwen-code/pull/8404
* fix(web-shell): discard stale automatic recaps by @carffuca in https://github.com/QwenLM/qwen-code/pull/8573
* fix(core): fail closed on zero inode file cache by @xianjianlf2 in https://github.com/QwenLM/qwen-code/pull/8290
* feat(review): say so when the bundle is older than the review it runs by @wenshao in https://github.com/QwenLM/qwen-code/pull/8390
* fix(core): bound backward transcript pages in long single-turn sessions by @wenshao in https://github.com/QwenLM/qwen-code/pull/8553
* feat(serve): observe daemon and child memory against real denominators by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/8423
* feat (web-shell): throttle Markdown AST parsing during streaming by @PratikWayase in https://github.com/QwenLM/qwen-code/pull/7904
* fix(core): report signal-terminated shell commands as errors by @daleselaji-dev in https://github.com/QwenLM/qwen-code/pull/8501
* feat(core): share compression caches with OpenAI providers by @DragonnZhang in https://github.com/QwenLM/qwen-code/pull/8418
* fix(ci): prevent npm cache producer from poisoning runners by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8669
* fix(serve): Allow approved same-host text reads outside workspace by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/8620
* fix(autofix): skip the PAT-backed jobs when the run has no secrets by @wenshao in https://github.com/QwenLM/qwen-code/pull/8671
* fix(triage): finalize the status comment on cancellation too by @wenshao in https://github.com/QwenLM/qwen-code/pull/8436
* feat(core): share compression cache with Gemini and Vertex AI by @DragonnZhang in https://github.com/QwenLM/qwen-code/pull/8425
* feat(channels): support group pairing by @qqqys in https://github.com/QwenLM/qwen-code/pull/8440
* fix(cli): preserve slash command names in narrow terminals by @carffuca in https://github.com/QwenLM/qwen-code/pull/8657
* feat(review): add qwen-code repository context manifest by @wenshao in https://github.com/QwenLM/qwen-code/pull/8654
* feat(dingtalk): include non-bot mention target identifiers in inbound context by @BenGuanRan in https://github.com/QwenLM/qwen-code/pull/8639
* fix(tests): avoid blocking the Vitest worker during directory E2E by @wenshao in https://github.com/QwenLM/qwen-code/pull/8685
* fix(cli): time out silent MCP SSE startup by @daleselaji-dev in https://github.com/QwenLM/qwen-code/pull/8555
* fix(channels): manage DingTalk interactive card config by @BenGuanRan in https://github.com/QwenLM/qwen-code/pull/8517
* perf(autofix): stop restoring a 2.65 GB npm cache to protect a 29 s install by @wenshao in https://github.com/QwenLM/qwen-code/pull/8681
* fix(ci): give review runs upstream-timeout headroom by @wenshao in https://github.com/QwenLM/qwen-code/pull/8673
* feat(autofix): bridge fork-PR reviews into the credentialed review lane by @wenshao in https://github.com/QwenLM/qwen-code/pull/8676
* fix(desktop): strip Windows verbatim prefix from workspace paths by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8619
* fix(desktop): fall back to system browser when built-in browser fails to open a link by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8594
* fix(cli): Run ACP agent fan-outs concurrently and past the tool-call cap by @wenshao in https://github.com/QwenLM/qwen-code/pull/8631
* feat(feishu): enrich observed contact labels by @BenGuanRan in https://github.com/QwenLM/qwen-code/pull/8569
* feat(web-shell): improve parallel agent activity feedback by @carffuca in https://github.com/QwenLM/qwen-code/pull/8559
* feat(external-context): Add optional Mem0 memory writes by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/8507
* fix(web-shell): keep mobile composer at chat pane bottom by @dreamWB in https://github.com/QwenLM/qwen-code/pull/8601
* fix(core): record the delivered prefix when a transport cut is continued by @harjothkhara in https://github.com/QwenLM/qwen-code/pull/8624
* perf(ci): run docs-only automatic reviews at medium effort by @wenshao in https://github.com/QwenLM/qwen-code/pull/8648
* fix(memory): Refresh live instructions after memory writes by @ZijianZhang989 in https://github.com/QwenLM/qwen-code/pull/8640
* feat(telemetry): attribute daemon-spawned sessions via channel (daemon/desktop) by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8670
## New Contributors
* @daleselaji-dev made their first contribution in https://github.com/QwenLM/qwen-code/pull/8501
**Full Changelog**: https://github.com/QwenLM/qwen-code/compare/v0.21.7...v0.21.7-nightly.20260808.4ec0371e6