Agent Breaking Changes_

← Qwen Code

Qwen Code Release v0.25.3-nightly.20261019.5dcf7c7cb1

modelpre

2026-10-19 · official notes

Summary

Full notes

<!-- Release notes generated using configuration in .github/release.yml at release/v0.25.3-nightly.20261019.5dcf7c7cb1 -->

## What's Changed
* feat(daemon): support session-targeted Skill and Extension refresh by @callmeYe in https://github.com/QwenLM/qwen-code/pull/14050
* fix(managed-agent): bound initial publication guard waits by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14563
* feat(runtime-broker): Session-create warm for Workspace-isolated bound Sessions by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14599
* test(cli): give turn-settling waits the slow-host budget by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14598
* feat(managed-agent): ship a configuration template in the reference distribution by @wenshao in https://github.com/QwenLM/qwen-code/pull/14631
* feat(serve): run private CSI file turns through Hosted Harness by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14487
* test(cli): retry ENOTEMPTY in hosted harness teardowns by @ZedingZhang in https://github.com/QwenLM/qwen-code/pull/14636
* feat(managed-agent): support local and SSH workspace file references by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14639
* test(cli): give the hosted probe-start guards a scheduling budget by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14662
* fix(managed-agent): recover committed wakes after writer loss by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14603
* feat(managed-agent): add a read-only --check preflight by @wenshao in https://github.com/QwenLM/qwen-code/pull/14632
* fix(sdk): size session create from the daemon session budget by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14588
* fix(serve): contain Hosted stdio MCP servers in a process group (#14503) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14640
* test(managed-agent): drain a held async Hook before the lifecycle Hooks on DELETE by @wenshao in https://github.com/QwenLM/qwen-code/pull/14638
* feat(serve): recover receipt-complete private CSI file turns by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14523
* docs(managed-agent): propose bounded retention for session state by @wenshao in https://github.com/QwenLM/qwen-code/pull/14659
* feat(managed-agent): raise attachment input budget to 8 MiB by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14628
* fix(managed-agent): recover the journal after a lapsed writer lease by @wenshao in https://github.com/QwenLM/qwen-code/pull/14570
* perf(managed-agent): index tool receipts by execution call by @wenshao in https://github.com/QwenLM/qwen-code/pull/14660
* fix(managed-agent): keep refused PreToolUse replacements out of Hosted after Hooks by @qqqys in https://github.com/QwenLM/qwen-code/pull/14650
* test(managed-agent): gate Spring crashes and the scanner lease for automation (FG11c slice 2) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14649
* feat(serve): read Turn recovery evidence from the Broker (D7-2) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14669
* fix(ci): repair Java gates and report failures on PRs by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14522
* fix(hosted): continue recovered background wake turns by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14679
* test(web-shell): widen Playwright wait budgets for shared CI hosts by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14643
* perf(web-shell): incrementally project managed session messages by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14562
* feat(managed-agent): ship systemd units for the single-host topology by @wenshao in https://github.com/QwenLM/qwen-code/pull/14633
* refactor(managed-agent): share CSI runtime marker reads by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14571
* fix(managed-agent): load saved tool details on expansion by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14604
* feat(managed-agent): park a Turn whose Runtime evidence stays unknown (D7-2) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14681
* test(managed-agent): drop the FG13a hook-catalog reply before registration by @wenshao in https://github.com/QwenLM/qwen-code/pull/14668
* fix(web-shell): preserve focus when trajectory execution views disappear by @qqqys in https://github.com/QwenLM/qwen-code/pull/14641
* feat(artifacts): record files the agent creates or edits by @zjgzx1988 in https://github.com/QwenLM/qwen-code/pull/14292
* fix(cli): reconcile managed recovery failure reasons by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14567
* test(managed-agent): assert G3 approval identities and pin drift by @wenshao in https://github.com/QwenLM/qwen-code/pull/14673
* fix(hooks): reap a Hook command unit when its worker dies (#14502) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14675
* test(managed-agent): drop the FG13a HTTP Hook reply after the request by @wenshao in https://github.com/QwenLM/qwen-code/pull/14692
* test(managed-agent): scope the FG11c crash assertions and fix the race fixture by @wenshao in https://github.com/QwenLM/qwen-code/pull/14682
* perf(managed-agent): bound the held-Runtime inventory read cost (#13930) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14691
* fix(config): let a Managed host restore its own session by @wenshao in https://github.com/QwenLM/qwen-code/pull/14688
* test(integration): widen workflow completion reply budgets past loaded-shard stalls (#14191) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/14202
* feat(managed-agent): bootstrap the shared deployment credentials by @wenshao in https://github.com/QwenLM/qwen-code/pull/14634
* feat(runtime): complete private CSI cold recovery by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14695
* fix(managed-agent): never fail a pending cancel over recovery (D7-2) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14694
* test(managed-agent): pin the drain flag on the release wire (#13930) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14712
* test(core): pin the Goal record's type/subtype gate by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14705
* test(web-shell): remove fixed transcript bundle size ceiling by @ytahdn in https://github.com/QwenLM/qwen-code/pull/14703
* fix(managed-agent): bound publication placement waits by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14542
* fix(managed-agent): Align image attachment formats with Qwen Code by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14704
* feat(extensions): add bounded generated HTML and JavaScript views by @qqqys in https://github.com/QwenLM/qwen-code/pull/14338
* docs(vscode): explain why webviews cannot use the same-origin exception by @Jun516 in https://github.com/QwenLM/qwen-code/pull/14684
* refactor(managed-agent): organize server packages by responsibility by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14710
* test(managed-agent): stabilize the FG12b released store fault gate (#14697) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/14711
* test(managed-agent): add G3 approval fault-point E2E arms by @wenshao in https://github.com/QwenLM/qwen-code/pull/14687
* test(core): synchronize scheduler disposal fault injection by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14727
* fix(ci): permit Java failure summaries on pull requests by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14717
* fix(managed-agent): reject an incomplete Workspace tool path in --check by @wenshao in https://github.com/QwenLM/qwen-code/pull/14635
* feat(workflow): authorize daemon ancestor workflows by @qqqys in https://github.com/QwenLM/qwen-code/pull/14646
* fix(cli): hold pre-runtime /acp WebSocket upgrades instead of falling through to a 406 (#13995) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/13997
* test(managed-agent): replay a denied Hook decision after a lost reply by @wenshao in https://github.com/QwenLM/qwen-code/pull/14701
* fix: align DeepSeek reasoning controls across official and MaaS routes by @qqqys in https://github.com/QwenLM/qwen-code/pull/14689
* fix(web-shell): surface the fix witness in the review artifact viewer by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14725
* feat(web-shell): show per-model TTFT and output speed in Token Usage by @qqqys in https://github.com/QwenLM/qwen-code/pull/14728
* fix(managed-agent): align fault gates with unsettled-append recovery by @yiliang114 in https://github.com/QwenLM/qwen-code/pull/14716
* test(cli): pin the FG13d pre-dispatch cancel in a unit test by @wenshao in https://github.com/QwenLM/qwen-code/pull/14735
* test(managed-agent): witness the durable hold sweep across both families (#13930) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14719
* test(managed-agent): widen the hosted recovery and hook-cancel gate windows (#14729) by @qwen-code-dev-bot in https://github.com/QwenLM/qwen-code/pull/14741
* feat(runtime): add original CSI application retirement by @doudouOUC in https://github.com/QwenLM/qwen-code/pull/14698
* feat(telemetry): Record reasoning effort on LLM spans by @zonemeen in https://github.com/QwenLM/qwen-code/pull/14718
* test(managed-agent): gate a disabled automation span (FG11d) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14731
* test(managed-agent): finish the FG13f decision-delivery arms by @wenshao in https://github.com/QwenLM/qwen-code/pull/14736
* test(managed-agent): leave a v2 execution unknown across a Harness kill (D7-2) by @wenshao in https://github.com/QwenLM/qwen-code/pull/14750
* docs(managed-agent): name the pre-dispatch commit-spy assertion by @wenshao in https://github.com/QwenLM/qwen-code/pull/14756

## New Contributors
* @Jun516 made their first contribution in https://github.com/QwenLM/qwen-code/pull/14684

**Full Changelog**: https://github.com/QwenLM/qwen-code/compare/v0.25.3...v0.25.3-nightly.20261019.5dcf7c7cb1
Weekly digest: every breaking change, retirement and price change across 23 coding agents, one email.

Form not loading? Subscribe here.