• 划词助手：支持文本选择快捷键、开关快捷键、思考块支持和引用功能
• 复制功能：新增纯文本复制（去除Markdown格式符号）
• 知识库：支持设置向量维度，修复Ollama分数错误和维度编辑问题
• 多语言：增加模型名称多语言提示和翻译源语言手动选择
• 文件管理：修复主题/消息删除时文件未清理问题，优化文件选择流程
• 模型：修复Gemini模型推理预算、Voyage AI嵌入问题和DeepSeek翻译模型更新
• 图像功能：统一图片查看器，支持Base64图片渲染，修复图片预览相关问题
• UI：实现标签折叠/拖拽排序，修复气泡溢出，增加引文索引显示

• fix(SelectionAssistant): ignore CtrlKey mode when ctrl+click
• fix(SelectionToolbar): prevent CSS updates in settings page
• fix(OpenAIProvider): ensure tool_calls are only yielded when present
• feat: enable rendering and download of inline base64-encoded images
• refactor(CodePreview): improve the triggering timing for highlighting
• perf: improve responsiveness on streaming formulas
• hotfix: gemini-2.5-pro-preview-06-05 using error
• fix(Inputbar): fix textarea expansion and collapse issues with long text (CherryHQ)
• fix(SelectionAssistant): support selection when alt key pressed
• refactor(SelectionToolbar): add transition effects to action buttons
• fix(SelectionToolbar): prevent dragging the demo
• fix(SelectionAssistant): default disabled
• fix(Inputbar): remove unnecessary flex properties from Inputbar styles
• hotfix: enhance OpenAI stream handling and error management
• feat: add set feed url functionality for early access
• feat(SelectionAssistant): add the "quote" action
• refactor: unified image viewer with integrated context menu
• fix: message editor doesn't resize
• fix: set thinking budget to 0 for gemini-2.5-flash when reasoning effort is off
• fix: couldn't edit text when sent file-only message
• fix: set message translate dropdown height
• fix: update silicon docs and models
• feat(restoreFromWebdav): make credentials and path optional
• fix: voyage ai can't be used on text embedding
• fix: update default translate model to deepseek-v3
• refactor: better semantic of obsidian export options
• 增加自动更新文档中版本号的github workflow
• Feature/dmxapi_images_to_image
• fix: prevent emoji picker from closing unexpectedly with IME
• fix: Implement label folding, drag-and-drop sorting of assistants within labels, and drag-and-drop sorting of labels
• fix(BackupManager): add content length to WebDAV file upload options
• fix(migrate): old translateModel incorrect
• fix: streamline file selection and ensure deletion of topic-related f…
• fix: update README files to enhance navigation and add project badges
• feat: add prompt variables description
• fix: ollama embedding knowledge query score always 100%
• fix: add Youdao and Nomic logos to model logo mapping
• fix: check if embedding is base64 encoded before convert it to float …
• feat: enhance unresponsive renderer handling and crash reporting
• refactor: use CodeEditor for customizing css
• feat(SelectionAssistant): support thinking block in action window
• fix: fix waring in usetags
• fix: shouldn't edit embedding dimension on existing knowledge base
• 翻译功能增加手动选择源语言的选项
• Fix: bubble-style unnecessary menu background (Plan D)
• feat: add citation index to show
• fix: bubble overflow patch
• fix(SelectionAssistant): reduce Copy conflict
• feat(i18n): add tooltips for model name in multiple languages
• Feat: Allows setting the vector dimension of the knowledge base embedding model
• fix(SelectionAssistant): improve auto-scroll behavior in action window
• feat(SelectionAssistant): shortcut key to toggle on/off
• fix: readme twitter link error
• feat: add plain text copy functionality for messages and topics. 添加了复制纯文本的功能（去除Markdown格式符号）
• feat(SelectionAssistant): add shortcut for selecting text
• fix(SelectionAssistant): shortcut in mac and running handling

• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in

Full Changelog:

We've packed v1.4.2 with usability upgrades, polished features, and stability enhancements. Check out what's new:

• Smarter Webapp Inputs: Automatically fill hidden fields using URL parameters ,

• Efficient Document Extraction: Faster, streamlined processing for Excel and CSV files , thanks to

• Marketplace UI Update: Easier browsing with improved sorting for app types , courtesy of

• Personalized Chatbot UX: Chat input placeholder now dynamically displays your bot's name , thanks to

• Robust File Uploads: Addressed issues with multiple file extension mappings ,

• Advanced Knowledge Base API Features:

  • Refined metadata filtering for precise querying. Targeted searches using chunk IDs ,
  • Improved tagging and dataset binding capabilities , thanks to

• Weave Tracing & W&B Integration: Enhanced support for dedicated cloud instances for better operations monitoring , from

• Flask-Cors Upgrade: Enhanced security by updating to the latest Flask-Cors version ,

• Resolved 401 Errors:

  • Smooth parallel workflow execution .
  • Fixed authentication issues in workflow_as_tool .
  • Eliminated backward invoke errors . All thanks to

• Agent Node Refinements:

  • Corrected enum syntax for Python 3.11 ,
  • Improved handling of long tokens in LLM invocation ,

• Annotation API Fix: Restored the missing end_user argument , courtesy of

• Docker Environment Corrections: Fixed erroneous handling of environment variables ,

• Code Node & Trace Management:

  • Enhanced array validation in code nodes ,
  • app_id now included in TraceTasks for better management , thanks to

• HTTP Node Reliability: Corrected curl operations using --data ,

• Memory Leak Prevention: Fixed Celery worker leaks ensuring jobs close correctly ,

• Improved Housekeeping Command: Now preserves avatar images and app icons correctly , thanks to

• Email Invitation Fixes: SMTP authentication errors resolved ,

• Consistent App Tagging: Resolved app tag update issues ,

• Markdown Rendering Fixes: Ensured abbreviations display properly in react-markdown ,

• LLM Node Image Parsing: Improved reliability for single-step executions with images ,

• Restored Agent Moderation: Functionality fully restored ,

• Plugin Extension Performance: Restored responsiveness and speed ,

• Question Classifier Stability: Executions now error-free , thanks to

• Enhanced Web UX: Single-run modals now auto-dismiss for better usability ,

Enjoy these improvements and a smoother, more stable experience! 🚀

1. Back up your customized docker-compose YAML file (optional)

   cd docker
   cp docker-compose.yaml docker-compose.yaml.$(date +%s).bak

2. Get the latest code from the main branch

   git checkout main
   git pull origin main

3. Stop the service. Please execute in the docker directory

   docker compose down

4. Back up data

   tar -cvf volumes-$(date +%s).tgz volumes

5. Upgrade services

   docker compose up -d

1. Stop the API server, Worker, and Web frontend Server.

2. Get the latest code from the release branch:

   git checkout 1.4.2

3. Update Python dependencies:

   cd api
   uv sync

4. Then, let's run the migration script:

   uv run flask db upgrade

5. Finally, run the API server, Worker, and Web frontend Server again.

• update img
• Chore/update img
• chore: enchance the copywriting of tool
• fix: i18n auto run failed
• fix: Enhances tenant ID handling in telemetry
• [Observability] Add type check and try-except in otel
• fix(workflow): fetch user failed when workflow run in parallel mode
• fix: Instance is not bound to a Session
• fix: reset password page dark style
• Fixes some i18n(ko) translations.
• fix(http): force multipart/form-data even without files
• fix: workflow plugins list update
• feat(agent_node): ensure that the enum-checking syntax is compatible with Python 3.11.
• fix: register user model to current_user in backward invoke.
• Fix/branding broken
• fix: inner invoke llm token too long
• chore: remove agent turn limits
• docs: Update PR template to emphasize guidelines and issue linking
• fix: show 'reset brand' button after set branding image
• fix: apps/annotation missing 1 required positional argument: 'end_user'
• fix: wrong env usage in middleware
• chore: improve error logging for requests to plugin daemon
• fix: handle values in output arrays for CodeNode transformation
• tests: Removes outdated marketplace download test
• refactor: Remove db from cycle manager
• refactor(workflow): Rename NodeRunMetadataKey to WorkflowNodeExecutionMetadataKey
• refactor(workflow): Rename workflow node execution models
• fix(models): WorkflowRun's total_steps and exceptions_count mismatch with database
• fix(ops_trace_manager): Adds app_id to TraceTask initialization
• chore: Colorize new OpenAI LLM versions
• Feat/15534 support replacing the bot in chat input placeholder with the bots name
• fix: import from curl not work for --data
• chore: translate i18n files
• refactor(api/core/workflow/enums): Rename WORKFLOW_RUN_ID to WORKFLOW_EXECUTION_ID
• fix: some display error in dark mode
• Refactor/markdown component split
• fix(json-schema-editor): Add container reference for resize observer in CodeEditor; Update language hook and help doc URL in JsonSchemaConfig
• fix: drop some type fixme
• Refactor/message cycle manage and knowledge retrieval
• fix celery job not closed issue
• Improve CONVERSATION_TITLE_PROMPT to correctly handle Japanese and input
• update knowledge base api
• Add APIs for Knowledge Base Tag Management and Dataset Binding
• fix: resolve unstable scrolling in workflow debug panel with multiple input fields
• fix: agent app tool update
• Amend color typo
• fix: fetch tenant_id in other trace providers besides langfuse
• fix: the plugin order is not the same as passed to api in DSL
• fix(housekeeping): exclude files that are used as app icons or avatar images from being removed
• nacos config init , and force add ts parms.
• fix ts5097
• check zilliz cloud of full-text search
• Fixes : Allow $ref in parameter for custom tools
• Fix : Force header in custom tool be string
• fix: ensure proper conversation role alternation for vLLM
• fix: When elasticsearch is used as the vector database, the Retrieval Test fails to filter the data after setting the Score Threshold, and the score of the recalled results is empty
• Fix/dark theme style issues
• fix: agent thought replaced by response text
• chore: update pnpm version to 10.11.1
• refactor: Replaces direct DB session usage with context managers
• fix: Upgrade Flask-Cors
• refactor: Removes unused LLMMode value_of method
• fix: Ensure model config integrity in retrieval processes
• fixes
• ♻️ refactor(middleware): remove duplicate CSP header assignment
• chore: prepare the plugin daemon base url to yarl URL ahead intstead of in every invocation
• Revert "♻️ refactor(middleware): remove duplicate CSP header assignment"
• fix: variable aggregator with group and file raise exception
• fix: ensure newlines around think tags for proper markdown rendering
• refactor: Removes tenant ID check from rate limit logic
• fix: unable to upload custom file in case of incorrect inffered by multiple extensions mapped from mime type with filename extension hints
• fix: adjust sticky header properties in Container component
• fix: update app tag error
• fix: autocorrect everything in web
• raise error when process_rule is required but missing
• chore: fix some security issues in markdown
• fix: plugin update redcorner mark display incorrect
• chore: ensure web code consistency by applying pnpm fix
• feat: plugin storage support volcengine tos
• fix(markdown): Ensure abbr: links render correctly in react-markdown v9+
• Feat/queue monitor
• refactor: Improve model status handling and structured output
• feat: allow fill inputs from url params
• assign dataset indexing_technique to args if not explicitly provided
• fix(llm_node): update file variable mapping to use vision configs
• Fix 500 error
• Fix/webapp access scope
• fix: the locale format
• Update template.zh.mdx-fix document update metadata body param
• Add vscode debugger
• chore: translate i18n files
• fix: update text_to_audio method to send data as JSON
• fix: agent moderation not working
• A more concise and effective extractor for excel and csv files
• Fix: style of radio checked
• feat: reorder app types
• Fix builtin_providers for tools.
• fix: opensearch fulltext search with metadata filtering dsl error
• fix(inner_api/plugin/wraps): refresh user model after creation in get user function
• chore: bump uv to 0.7.x
• fix: opensearch metadata filtering returns empty
• chore: remove repeat public api and service api panel
• chore: replace pseudo-random generators with secrets module
• feat: add browser list
• chore: chart panel ui enhance
• fix: opensearch vector search falls back to keyword search
• fix: missing bot name in orchestrate
• feat(api): Adjust WorkflowDraftVariable and WorkflowNodeExecutionModel
• Add support for W&B dedicated cloud instances in Weave tracing integration
• fix: clean up two unreachable code
• Fix/webapp no permission page
• fix: some dark mode display incorrect
• Feat/webapp verified sso main
• refactor(DSL imports): using organization/name/version to fetch DSL dependencies.
• fix(api): Resolve error encountered when executing QuestionClassifieNode
• refactor(api): Decouple ParameterExtractorNode from LLMNode
• fix(web): optimize prompt change logic for LLM nodes
• fix auto metadata filter
• chore: update plugin publish link text
• chore: bump mypy to 1.16
• fix: only enterprise version request app access mode
• refactor: replace compact response generation with length-prefixed response for backwards invocation api
• chore(package): Bump version to 1.4.2

• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in

Full Changelog:

合并了佬的一个PR，可以看到订阅链接的失败情况，失败节点超过80%就会打印日志出来

subs-check 问题沟通、使用交流渠道：

• 3f4cc46140af632fb8cb7114e1ad48da04ff800d feat(proxy): 为节点添加订阅链接来源并检查成功率
• 5edbb999f03550bcbb188de6d4568b84f1960496 fix
• dadc7075d584d2941b84a7d6dfb2fbcb0053d2e8 op: add get debug log

• feat(ai-proxy): support Amazon Bedrock Image Generation
• fix: Fix the incorrect rewrite config generated for Nacos 3 MCP Servers
• feat: update translate-readme action
• fix: Remove the Authorization request header when using AI-proxy to proxy Gemini
• fix : fix issue
• Add test translation workflow
• mcp server support API auth through OAS3 security schemes
• Fix : add fail strategy for wasmplugin generated by mcp server
• fix: modify log level WARN -> DEBUG in key-auth plugin
• fix proxy-wasm-cpp-sdk
• feat: Supports recording request header, request body, response header and response body information in the access log
• feat(mcp-server): add HackMD mcp server
• add mcp service shebao tools
• feat: Add a github action to copy CRD definitions from api folder to helm folder
• feat: allow skipping higress dev image build during wasmplugin e2e tests
• feat: Refactor mcpServer.matchList config generation logic
• add info log of ai-search plugin
• fix(ai-proxy): URL encode model name in Bedrock requests
• feat(ai-proxy): add doubao Image Generation support
• feat: cluster-key-rate-limit support setting global rate limit thresholds for routes​
• feat(ai-proxy): support OpenAI-compatible image and audio model Mapping
• fix: set "EnableSemanticCachefalse" to false when no vector configured in ai-cache
• feat(ai-proxy): add batches & files support
• feat: support dify ai-proxy e2e test || feat: support diify ai-proxy e2e test
• fix content-length header not remove in ai-search plugin
• feat(ai-proxy): add modelMapping regexp support
• feat(ai-proxy): Fixed the issue that the API pass-through path error does not support openaiCustomUrl after openai is configured.
• feat(frontend-gray): Add uniqueGrayTag configuration detection
• feat(ai-proxy): add models & image generation support for gemini
• feat(ai-proxy): support Google Cloud Vertex
• add upstream override wasm abi
• feat(ai-proxy): Add Claude image understanding and Tools calling capabilities
• fix: refactored mcp server auto discovery logic and fix some issue
• fix : fix credential process logic for nacos mcp util and add ut for it
• fix: Support mixing line breaks in a single SSE response
• Update CRD file in the helm folder

• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in
• made their first contribution in

Full Changelog:

This release introduces parallel node features, UI optimizations, file extraction improvements, and critical bug fixes for stability and usability.

• [elsa] 并行节点特性添加
• feat: Add readOnly mode to JadeFlowEntry for history view

• [elsa] 大模型节点模型下拉框宽度自适应
• [elsa] elsa-react打包添加翻译文件
• [elsa] 添加README
• [elsa] 知识检索节点参数修改
• [elsa] enhance reference-type config protection

• [elsa] 序列化保存改同步
• [elsa] 隐藏文件提取节点中的提取说明配置,避免引起歧义
• [elsa] Bug Fixes: Address UI Dropdown Resize, Knowledge Graph Updates, and GraphOperator.js Stability
• [elsa] ci: fix version branch regex matching
• [elsa] fix(GraphOperator): optimize single-key update handling
• fix(ConnectorProvider): Sync connector name changes with connected lines
• [elsa] fix(OutputVariableRow): preserve data type when switching from Reference to Input
• [elsa] fix(JadeObservableOutput): conditionally render description column to avoid empty space
• [elsa] fix(JadeReferenceTreeSelect): Force reference updates on same selection

Full Changelog: elsa-v0.1.0...elsa-v0.2.0

• Parallel Node Support : Added parallel execution capabilities for workflow nodes.
• Dynamic UI Adjustments : Model selection dropdown now auto-adjusts width for better readability.
• Localization Support : Added translation files for elsa-react builds.
• Knowledge Retrieval Node Updates : Modified parameters for improved functionality.
• Read-Only Mode for Flow History : Enabled non-editable view for historical workflow inspection.

• Synchronous Serialization : Changed serialization to synchronous to prevent state corruption.
• File Extraction Clarity : Removed ambiguous extraction instructions from file nodes.
• Connector Sync Fix : Ensured connector name changes propagate to connected lines.
• Data Type Preservation : Fixed data type loss when switching variable modes.
• UI Optimization : Eliminated empty space in description columns.
• Reference Update Reliability : Forced updates on repeated selections in JadeReferenceTreeSelect.

• Added README : Improved project documentation for better onboarding.

• Breaking Change: Removed deprecated userId parameter in favor of knowledgeConfigId for improved traceability.
• Added DEFAULT_KNOWLEDGE_RETRIEVAL_NODE_KNOWLEDGE_CONFIG_ID constant.
• Upgraded UpdateGroupIdReducer to UpdateGroupIdAndConfigIdReducer for synchronized updates of groupId and knowledgeConfigId.

• Model Selection Dropdown: Now dynamically adjusts width based on content length via dropdownMatchSelectWidth={false}.

• Auto-Creation: The update method now automatically initializes missing config paths.
• Type Inference: New getTypeFromUpdates utility infers data types (String, Array, Object, etc.) from input.

• Branch Matching: Fixed regex to correctly target version branches (e.g., elsa-0.1.x). Escaped decimal points in patterns.

• Migration Required: Existing flows usinguserIdmust switch toknowledgeConfigId.
• CI/CD Compliance: Version branches must follow elsa-{major}.{minor}.x format (e.g., elsa-0.1.x).
• Serialization Behavior : Now synchronous—ensure dependent code handles blocking operations.
• File Extraction Node : Extraction instructions removed; update any dependent configurations.

Special thanks to all contributors for this release:

• Select custom locations to check for apps
• Latest now clearly distinguishes between apps with full support and those with limited support. The latter are not shown by default, but can be enabled in Settings.
• Added an Open action to the File menu. (thanks

• Fixed a hang when opening other apps from Latest (thanks
• Fixed a crash when open Latest without having a working internet connection
• Attempt to fix issues with Latest showing no apps at all
• Ensures Latest correctly restores its window size and position
• Fixed a glitchy progress indicator on macOS 15.4 and later
• Various interface fixes
• Various localization updates (thanks to all the contributors!)

Please note that the minimum supported OS version is now macOS 10.15 Catalina.