Qwen Code v0.21.1-nightly.20260731.702932cc7 预发布

<!-- Release notes generated using configuration in .github/release.yml at release/v0.21.1-nightly.20260731.702932cc7 --> ## What's Changed * fix(ci): add default bash shell to container jobs in qwen-triage by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7838 * fix(web-shell): preserve pasted text in composer by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7824 * feat(web-shell): add monitor task details by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7817 * fix(ci): add git safe.directory for container jobs by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7843 * fix(ci): add --init to container jobs to reap zombie processes by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7848 * fix(core): wait for output stream flush before settling background shells by @ComplexSimply in https://github.com/QwenLM/qwen-code/pull/7833 * fix(scripts): retry model calls and surface degraded release notes by @he-yufeng in https://github.com/QwenLM/qwen-code/pull/7535 * feat(web-shell): Scope voice to composer workspace by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7754 * feat(hooks): Add submitted prompt provenance by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7762 * fix(ci): prevent worktree cleanup from failing the verify job by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7857 * perf(acp): Preload providers after session creation by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7767 * test(channels): run the feishu, weixin and qqbot suites in CI by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7853 * feat(autofix): retry deterministic rejection once by @qqqys in https://github.com/QwenLM/qwen-code/pull/7796 * fix(release): publish all channel packages, not just channel-base by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7845 * feat(web-shell): add Channel management page by @qqqys in https://github.com/QwenLM/qwen-code/pull/7793 * feat(acp): add session-scoped runtime MCP by @qqqys in https://github.com/QwenLM/qwen-code/pull/7847 * feat(core): persist and replay Goal v3 state by @qqqys in https://github.com/QwenLM/qwen-code/pull/7815 * feat(core): add full-resolution image zoom tool by @qqqys in https://github.com/QwenLM/qwen-code/pull/7809 * fix(scripts): harden retry classification and preserve deadline error context by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7854 * feat(web-shell): add composer footer renderer by @dreamWB in https://github.com/QwenLM/qwen-code/pull/7856 * fix(core): fast-fail permanent quota-exhaustion 429s instead of silent retry by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7842 * fix(web-shell): isolate history and session drafts by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7810 * fix(triage): retry a transient npm ci before blaming the PR for it by @wenshao in https://github.com/QwenLM/qwen-code/pull/7884 * fix(webui): fall back to history_truncated marker recordId for transcript pagination anchor by @wenshao in https://github.com/QwenLM/qwen-code/pull/7829 * feat: Gate session writer lease behind opt-in by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7894 * fix(review): give the review retry the remaining time budget by @wenshao in https://github.com/QwenLM/qwen-code/pull/7852 * fix(integration): configure Docker sandbox networking for submitted-prompt provenance test (#7879) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7881 * fix(cli): report a genuine $0.00 cost instead of N/A by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7784 * fix(core): reject socks5h and socks4a proxy URLs by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7786 * fix(core): correct the character classes in checkContentLoop by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7788 * fix(web-shell): make /copy with a bare index work by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7789 * fix(core): decline combined sed flags where -i is not last by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7790 * fix(core): pass the Grep pattern behind -e so a leading dash is not an option by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7863 * feat(ci): Deduplicate E2E failure issues by commenting on existing issue by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7792 * fix(core): cap a bare Error's message like every other getErrorMessage path by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7865 * fix(cli): make wrapToVisualLines count zero-width characters like its sibling by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7873 * fix(core): charge the separator and ellipsis to the preview budget by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7874 * fix(cli): do not count a partial trailing line when re-opening a split fence by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7875 * fix(cli): make /copy <message> <index> select the code block by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7883 * feat(triage): add revert-pattern high-risk path detection by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7414 * feat(web-shell): add native workspace folder picker by @qqqys in https://github.com/QwenLM/qwen-code/pull/7849 * feat(web-shell): add Channel configuration flows by @qqqys in https://github.com/QwenLM/qwen-code/pull/7893 * fix(test): Restore first-output benchmark measurement validity and correct its artifact schema by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7820 * fix(daemon): harden Todo Stop Guard continuations by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7821 * fix(core): read the stash reflog from the common git dir by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7774 * fix(core): keep Draft 4 boolean exclusive bounds in toOpenAPI30 by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7782 * fix(core): apply maxDepth to flat-format memory imports by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7851 * fix(core): render a thought part's reasoning instead of the boolean flag by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7866 * fix(core): bridge tool-result images for text-only models by @LaZzyMan in https://github.com/QwenLM/qwen-code/pull/7484 * fix(cli): patch ink to clear staticNode on indirect subtree removal by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7816 * fix(core): auto-retry transient network errors during API calls by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7898 * fix(cli): hide stale sticky todos from previous turns by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7900 * fix(serve): Release managed session writer locks on shutdown by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7812 * perf(core): add early Anthropic cache breakpoint on the stable system prefix by @DragonnZhang in https://github.com/QwenLM/qwen-code/pull/7912 * fix(cli): add polling fallback for git branch name display by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7830 * fix(cli): remove redundant 'Read file' prefix from @mention tool card by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7902 * feat(core): Add ARMS session user ID by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7921 * fix(safe-mode): preserve caller-supplied top-tier MCP servers by @VitaliBabkin in https://github.com/QwenLM/qwen-code/pull/7827 * fix(ci): keep the post-merge E2E signal on main alive by @wenshao in https://github.com/QwenLM/qwen-code/pull/7795 * fix(core): short-circuit the flush wait when the output stream cannot flush by @ComplexSimply in https://github.com/QwenLM/qwen-code/pull/7905 * fix(core): track quotes inside a command substitution in splitCommands by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7870 * fix(core): count the per-server and per-tool always-allow outcomes as approvals by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7869 * fix(web-shell): report intended workspace to host when starting a new chat by @wenshao in https://github.com/QwenLM/qwen-code/pull/7910 * fix(cli): default to virtualized terminal history by @ZevGit in https://github.com/QwenLM/qwen-code/pull/5738 * fix(ci): restore workspace ownership in cleanup to prevent EACCES by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7931 * perf(ci): cut the E2E suite from ~40min to ~24min by @wenshao in https://github.com/QwenLM/qwen-code/pull/7798 * feat(channels): expose loop tools in daemon sessions by @qqqys in https://github.com/QwenLM/qwen-code/pull/7891 * fix(review): recover the resolved effort when --effort is not re-threaded by @wenshao in https://github.com/QwenLM/qwen-code/pull/7855 * feat(web-shell): honor voice hold mode by @callmeYe in https://github.com/QwenLM/qwen-code/pull/7839 * feat(web-shell): manage Channel pairing requests by @qqqys in https://github.com/QwenLM/qwen-code/pull/7909 * feat(channels): dispatch GitHub notifications by reason by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7826 * feat(web-shell): add git branch picker, commit dialog, and create PR flow by @wenshao in https://github.com/QwenLM/qwen-code/pull/7731 * feat(triage): surface the sandboxed lanes on the CI path by @wenshao in https://github.com/QwenLM/qwen-code/pull/7917 * fix(scripts): slim release-note model prompts and log request timing by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7941 * feat(web-shell): suggest BTW for side questions by @carffuca in https://github.com/QwenLM/qwen-code/pull/7935 * fix(release): pin channel-base dep to exact version during release bump by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7953 * feat(triage): make the verify report readable in Chinese by @wenshao in https://github.com/QwenLM/qwen-code/pull/7918 * fix(triage): make the build-process guard diagnosable and zombie-aware by @wenshao in https://github.com/QwenLM/qwen-code/pull/7858 * fix(ci): give each job its own proxy wrapper directory by @wenshao in https://github.com/QwenLM/qwen-code/pull/7951 * feat(autofix): defer suggestions after five change rounds by @qqqys in https://github.com/QwenLM/qwen-code/pull/7913 * feat(core): connect Goal v3 to the turn engine by @qqqys in https://github.com/QwenLM/qwen-code/pull/7895 * feat(core): bound image reads for reliable zoom by @qqqys in https://github.com/QwenLM/qwen-code/pull/7911 * fix(integration): scale interactive read-then-write waits with env timeout (#7942) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7943 * test(integration): deflake asyncGenerator canUseTool content assertion (#7937) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7939 * fix(ci): clean stale .qwen/ before checkout to prevent Permission denied by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7977 * feat(triage): make the not-verified sentence a mechanical 2b-bis trigger by @wenshao in https://github.com/QwenLM/qwen-code/pull/7965 * fix(ci): restore verify workspace permissions by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7992 * feat(external-context): Add submitted-prompt auto recall by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7877 * fix(ci): update the Qwen binary used by ECS runners by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8000 * feat(web-shell): render streaming charts with markdown-chart by @zhangxy-zju in https://github.com/QwenLM/qwen-code/pull/7916 * feat: robust ripgrep by @water-in-stone in https://github.com/QwenLM/qwen-code/pull/7888 * fix(core): mark folders the item budget never expanded by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7868 * fix(feishu): keep code-fence chunks within the message size limit by @chinesepowered in https://github.com/QwenLM/qwen-code/pull/7850 * fix(web-shell): preserve token and base path in session URLs by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7926 * ci: add isolated DSW SWE-bench release pipeline by @DennisYu07 in https://github.com/QwenLM/qwen-code/pull/7656 * fix(web-shell): polish Channel pairing requests by @qqqys in https://github.com/QwenLM/qwen-code/pull/7997 * fix(cli): correct hardware cursor off-by-one in fullscreen mode by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7998 * fix(cli): prevent SGR mouse events from being swallowed as paste on Windows by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7988 * feat(channels): add DingTalk interactive cards by @BenGuanRan in https://github.com/QwenLM/qwen-code/pull/6930 * feat(web-shell): refine advanced table controls by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7999 * fix(serve): allow bounded reads of large text files by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7947 * fix(web-shell): prefer artifact type metadata in cards by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7973 * fix(web-shell): reduce composer input latency by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8015 * fix(webui): stabilize session history pagination by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8001 * fix(release): skip notes-start-tag when previous release diverges from target (#7969) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7970 * fix(sdk): raise daemon browser bundle budget to 177KB by @wenshao in https://github.com/QwenLM/qwen-code/pull/8024 * fix(cli): map Kitty Super (Command) modifier to meta by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7996 * fix(daemon): distinguish draining channel worker diagnostics by @destire-mio in https://github.com/QwenLM/qwen-code/pull/7932 * feat(triage): lead the verify comment with a qualitative verdict, fold the Chinese by @wenshao in https://github.com/QwenLM/qwen-code/pull/7974 * fix(review): preflight free disk before build-test installs and builds by @wenshao in https://github.com/QwenLM/qwen-code/pull/7986 * test(integration): migrate flaky E2E tests to fake-openai-server by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7934 * feat(triage): run external /verify on ECS behind a workspace wipe by @wenshao in https://github.com/QwenLM/qwen-code/pull/7985 * fix(cli): reserve Ctrl+Shift+C for terminal copy, not quit/clear by @chiga0 in https://github.com/QwenLM/qwen-code/pull/8011 * fix(core): auto-record artifact file writes by @chiga0 in https://github.com/QwenLM/qwen-code/pull/7914 * fix(cli): MCP prompt completion no longer blocks Enter for optional params by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/7995 * feat(hooks): add security.allowPrivateNetworkHooks to bypass SSRF range checks for trusted scopes by @xurik in https://github.com/QwenLM/qwen-code/pull/7968 * feat(web-shell): add split pane header action slot with overflow by @samuelhsin in https://github.com/QwenLM/qwen-code/pull/7808 * feat(channels): add GitLab polling channel adapter by @OrbitZore in https://github.com/QwenLM/qwen-code/pull/7862 * fix(channels): disable native cron in daemon sessions by @qqqys in https://github.com/QwenLM/qwen-code/pull/8034 * feat(channels): add pairing approval management API by @qqqys in https://github.com/QwenLM/qwen-code/pull/8045 * feat(github-channel): filter notification reasons by @xianjianlf2 in https://github.com/QwenLM/qwen-code/pull/8031 * fix(channels): reject unusable GitHub self-allowlists by @zjunothing in https://github.com/QwenLM/qwen-code/pull/8055 * fix(core): drop top-level oneOf from send_message tool schema by @netbrah in https://github.com/QwenLM/qwen-code/pull/7989 * fix(cli): stop resize repaint from causing scroll storm by @chiga0 in https://github.com/QwenLM/qwen-code/pull/8009 * feat(review): disclose a zero-finding Approve on a non-trivial diff as low signal by @wenshao in https://github.com/QwenLM/qwen-code/pull/7987 * fix(core): rebind fork capabilities on resume by @DragonnZhang in https://github.com/QwenLM/qwen-code/pull/7927 * feat(core): preload deferred tools within a context-window threshold by @DragonnZhang in https://github.com/QwenLM/qwen-code/pull/7922 * fix(channels): make GitHub final response publication single-shot by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8033 * fix(ci): restore workspace ownership before checkout in PR review job by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8062 * fix(release): bump preview base past published stable by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7978 * test(web-shell): expand restored-history pagination regression coverage by @PratikWayase in https://github.com/QwenLM/qwen-code/pull/7907 * fix(serve): Add certified session writer handoff by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7976 * test(integration): Measure immediate prompt dispatch stages by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7994 * fix(core): allow transport stream retry during the thinking-only phase by @ComplexSimply in https://github.com/QwenLM/qwen-code/pull/7938 * fix(github-channel): validate and document reasonFilter by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8035 * fix(core): guard against CJK-driven char/4 under-count in output clamp by @zambalee in https://github.com/QwenLM/qwen-code/pull/7963 * fix(core): preserve active Todo context across tool turns by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/7919 * feat(review): add `review run` — headless review with a machine-readable verdict by @wenshao in https://github.com/QwenLM/qwen-code/pull/7983 * fix(cli): stamp QWEN_CODE_CLI at the workspace entry and publish QWEN_CODE_MODEL by @wenshao in https://github.com/QwenLM/qwen-code/pull/7993 * feat(verify-pr): add seven techniques from maintainer verification rounds by @wenshao in https://github.com/QwenLM/qwen-code/pull/8010 * feat(triage): raise the /verify agent budget from 25m to 120m by @wenshao in https://github.com/QwenLM/qwen-code/pull/8014 * fix(autofix): cumulative timeout breaker, narrowed retry prompt, truthful handoff wording by @wenshao in https://github.com/QwenLM/qwen-code/pull/8044 * feat(autofix): post a takeover milestone digest every tenth pushed round by @wenshao in https://github.com/QwenLM/qwen-code/pull/8046 * fix(autofix): post the takeover engage ack from the command itself by @wenshao in https://github.com/QwenLM/qwen-code/pull/8043 * test(e2e): force delegation in flaky subagent case (#8070) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/8073 * fix(web-shell): stabilize enhanced table controls by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8041 * feat(github-channel): add transient working reaction by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/8061 * fix(integration): make interactive read-then-write test deterministic (#8060) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/8064 * feat(autofix): back off scan inspection of idle candidates by @wenshao in https://github.com/QwenLM/qwen-code/pull/8049 * fix(web-shell): show server queue status for pending messages by @wenshao in https://github.com/QwenLM/qwen-code/pull/8065 * fix(web-shell): improve artifact previews by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8078 * feat(triage): make /verify evidence screenshots actually possible by @wenshao in https://github.com/QwenLM/qwen-code/pull/8016 * fix(ci): gate the attachment guard before it allocates a runner by @wenshao in https://github.com/QwenLM/qwen-code/pull/8095 * fix(web-shell): translate legacy /skills invocations by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8103 * fix(verify-pr): budget evidence capture in scope selection by @wenshao in https://github.com/QwenLM/qwen-code/pull/8104 * fix(web-shell): show prompt send failure and retry action by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8106 * fix(web-shell): make question submission retryable by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8096 * fix(web-shell): preserve approved tool results by @ytahdn in https://github.com/QwenLM/qwen-code/pull/8099 * fix(autofix): answer round-cap refusals on the PR instead of only in logs by @wenshao in https://github.com/QwenLM/qwen-code/pull/8067 * fix(autofix): salvage race-lost pushes by merging the moved head and retrying by @wenshao in https://github.com/QwenLM/qwen-code/pull/8042 * feat(autofix): per-source feedback budget in Critical-only mode by @wenshao in https://github.com/QwenLM/qwen-code/pull/8071 * feat(core): tag UserPromptSubmit hook context and record display provenance by @zjgzx1988 in https://github.com/QwenLM/qwen-code/pull/7956 * fix(core): abort workflows during session shutdown by @qqqys in https://github.com/QwenLM/qwen-code/pull/8107 * fix(test): resolve turn completion on result messages in setModel E2E test (#8072) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/8075 * feat(serve): page large text files by byte cursor by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/8002 * fix(test): give multi-model E2E turns more CI timeout headroom (#8108) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/8111 * fix(serve): Prevent repeated workspace skill rescans by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/8080 * feat(review): statement-level mutation probes in test-efficacy by @wenshao in https://github.com/QwenLM/qwen-code/pull/8020 * feat(web-shell): add contextual task panels by @ytahdn in https://github.com/QwenLM/qwen-code/pull/7929 * feat(agent): add fork tool execution allowlist by @destire-mio in https://github.com/QwenLM/qwen-code/pull/8066 * fix(serve): Isolate daemon session maintenance writers by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/7975 * fix(serve): restore build by routing side-task rollback through the workspace session service by @wenshao in https://github.com/QwenLM/qwen-code/pull/8144 * fix(web-shell): isolate worktree session execution by @wenshao in https://github.com/QwenLM/qwen-code/pull/8068 * feat(web-shell): manage pairing approvals by @qqqys in https://github.com/QwenLM/qwen-code/pull/8081 * fix(test): allow slow permission mode E2E responses by @wenshao in https://github.com/QwenLM/qwen-code/pull/8149 * fix(test): make auto-edit permission E2E deterministic by @wenshao in https://github.com/QwenLM/qwen-code/pull/8154 ## New Contributors * @VitaliBabkin made their first contribution in https://github.com/QwenLM/qwen-code/pull/7827 * @xianjianlf2 made their first contribution in https://github.com/QwenLM/qwen-code/pull/8031 * @zambalee made their first contribution in https://github.com/QwenLM/qwen-code/pull/7963 **Full Changelog**: https://github.com/QwenLM/qwen-code/compare/dsw-manual-poc-20260727-1...v0.21.1-nightly.20260731.702932cc7