v0.4.13

What's Changed

• refactor(langchain): extract standalone integration package by @ehz0ah in #3685
• feat(bot): support streaming for OpenAI and VLM credentials by @yeshion23333 in #3696
• fix(resource): preserve synchronous ingestion errors by @qin-ctx in #3698
• feat(ragfs): add structured lock tracing logs by @baojun-zhang in #3693
• fix(core): harden privacy and background failure handling by @ehz0ah in #3548
• feat: 支持基于火山方舟的音视频多模态理解 by @Hao-Yu-la in #3563
• refactor(vectordb): unify random recall via client-generated vector by @zhoujh01 in #3706
• feat(config): add enable_watch_scheduler toggle for read-only replicas by @fengluodb in #3704
• fix(server): read attrs tags by exact URI and level by @zhoujh01 in #3707
• feat(agent-evolution): improve provenance, history, and account settings by @heaoxiang-ai in #3695
• fix(packaging): decouple LangChain client from server image by @t0saki in #3711
• feat(vectordb): add text field type for large strings by @zhoujh01 in #3725
• feat(integrations): add ZCode memory plugin by @TeFuirnever in #3678
• feat(queue): make ExternalParse worker concurrency configurable by @KCHENPENGFEI in #3710
• fix(zcode): make memory capture reliable by @t0saki in #3728
• feat(search): expose tags in find/search results and drop empty fields by @zhoujh01 in #3730
• feat(agent-evolution): track experience trajectory lineage by @heaoxiang-ai in #3727
• Fix Web Studio terminal panel crash by @yufeng201 in #3735
• 资源导入：支持解析后不拆分文档 by @KCHENPENGFEI in #3645
• feat(retrieval): assemble auto-recall context server-side via /search mode="context" by @t0saki in #3534
• feat(agent-evolution): aggregate outcomes and refine experience snapshots by @heaoxiang-ai in #3742
• Feat/session auto commit v2 by @zhoujh01 in #3736
• fix(compile): resolve pickle error and merge skill_name changes by @fujiajie666 in #3738
• fix(overview):language fix by @fujiajie666 in #3747
• feat(queuefs): support redis as queufs backend ,redis mode support singleton 、cluster 、 sentinel by @baojun-zhang in #3741
• feat: support processing mode for content write by @zhoujh01 in #3615
• fix(retrieval): honor context tier ceilings and stop cooling unserved recalls by @t0saki in #3746
• feat: OIDC and LDAP as new auth mode for OpenViking by @MaojiaSheng in #3708
• docs: update WeChat group QR code by @qin-ctx in #3819
• feat(queuefs): add bounded redis startup stale-recovery sweeps by @baojun-zhang in #3748
• feat(server): make HTTP keep-alive idle timeout configurable by @jiakun-lu in #3743
• fix(session): recover partial capture sessions by @t0saki in #3820
• fix(memory-plugin): preserve commit payload on retry by @t0saki in #3822
• fix(claude-plugin): count recalled memory URIs by @t0saki in #3821
• feat(openclaw): add Experience Memory tools by @heaoxiang-ai in #3827
• fix(openclaw-plugin): 修复 apiKey 类型导致的 release 构建失败 by @ZaynJarvis in #3832
• feat(ci): openclaw plugin 合入 main 自动发包 by @ZaynJarvis in #3836
• 优化 TaskTracker 并发，消除全局异步锁阻塞 by @KCHENPENGFEI in #3829
• fix(openclaw-plugin): 去掉 SecretRef exec，解除 ArkClaw 安装拦截 by @ZaynJarvis in #3849

New Contributors

• @TeFuirnever made their first contribution in #3678
• @jiakun-lu made their first contribution in #3743

Full Changelog: v0.4.12...v0.4.13