nightly

Release nightly created from c217b8f at 2026-03-06 21:26:49+08:00

v0.24.0

Summary

New features

• Memory
  • Introduces APIs and an SDK for developer integration.
  • Adds Memory extraction log display in the console for improved debugging and tracing.
• Dataset
  • Added support for batch management of Metadata.
  • Renamed "ToC (Table of Contents)" to "PageIndex".
• Agent
  • Launches a new Chat-like Agent conversation management interface that retains Sessions and dialogue history.
  • Introduces a multi-Sandbox mechanism, currently supporting local gVisor and Alibaba Cloud, with compatibility for mainstream Sandbox APIs (configurable in the Admin page).
• Chat
  • Adds a new "Thinking" mode and removed the previous "Reasoning" configuration option.
  • Optimizes retrieval strategies for deep-research scenarios, enhancing recall accuracy.
• Admin
  • Adds support for configuring multiple Admin accounts.
• Model configuration center
  • Adds a model connection test feature when adding new models.
• Ecosystem
  • Adds support for OceanBase as a database alternative to MySQL.
  • Adds support for PaddleOCR-VL.
• Model
  • Adds new model support for Kimi 2.5, Stepfun 3, and doubao-embedding-vision, among others.
• Data sources
  • Adds new data source integrations for Zendesk, Bitbucket, and others.

What's Changed

• Fix: KG search issue. by @KevinHuSh in #12364
• Feat: add Zendesk data source integration with configuration and sync capabilities by @buua436 in #12344
• Refactor docs of RAGFlow admin by @JinHai-CN in #12361
• Docs: Added v0.23.1 release notes by @writinwaters in #12371
• Fix urls for basic docs by @yingfeng in #12372
• Docs: Updated v0.23.0 release notes by @writinwaters in #12374
• Feat: Bitbucket connector by @Magicbook1108 in #12332
• Fix: metadata data synchronization issues; add memory tab in home page by @dcc123456 in #12368
• Fix: webdav connector by @Magicbook1108 in #12380
• Feat: Refactoring the documentation page using shadcn. #10427 by @cike8899 in #12376
• Fix Tika version mismatch in Dockerfile.deps (3.0.0 → 3.2.3) by @mmtmr in #12267
• Update README by @yingfeng in #12386
• ADMIN CLI: support grant/revoke user admin authorization by @JinHai-CN in #12381
• Update HELP command of ADMIN CLI by @JinHai-CN in #12387
• Feat: read web api testcases by @Lynn-Inf in #12383
• Fix: Fixed the issue where the upload DSL dialog box was too narrow. #10427 by @cike8899 in #12384
• refactor:re order logics in clean_gen_conf by @Woody-Hu in #12391
• Remove duplicated tag_feas assignment in create route (api/apps/chunk_app.py) by @kyuengmanKim in #12392
• security: harden OpenDAL SQL initialization against injection by @RinZ27 in #12393
• Fix: remvoe async wrappers by @Magicbook1108 in #12405
• Fix: Not able to add MCP Server #12394 by @dcc123456 in #12406
• Improve task executor heartbeat handling and cleanup by @maoxiaowang in #12390
• Test: get message content testcase by @Lynn-Inf in #12403
• feature: add system setting service by @JinHai-CN in #12408
• doc: update admin CLI by @JinHai-CN in #12413
• Feat: message write testcase by @Lynn-Inf in #12417
• Chore(compose): remove unnecessary history_data_agent volume mount by @asiroliu in #12418
• Refactor: UmiJs -> Vite by @dcc123456 in #12410
• Fix:bug fix by @buua436 in #12423
• Refactor setting type by @JinHai-CN in #12425
• Refactor: Refactoring VolcEngine and Yiyan modal using shadcn. #10427 by @cike8899 in #12426
• Feat: PDF vision figure parser supports reading context by @yongtenglei in #12416
• Refa: improve TOC building with better error handling by @asiroliu in #12427
• Fix: handle invalid img_id format in chunk update by @majiayu000 in #12422
• Add list configs and environments by @JinHai-CN in #12438
• Update admin doc by @JinHai-CN in #12439
• Refactor: Refactoring AzureOpenAIModal using shadcn. #10427 by @cike8899 in #12436
• Fix: Some bugs by @dcc123456 in #12441
• Feat: process memory by @Lynn-Inf in #12445
• Refactor: Refactor FishAudioModal and BedrockModal using shadcn. #1036 by @cike8899 in #12449
• fix: add 'mom_id' column to OBConnection chunk table by @whhe in #12444
• Fix: add uv binary archive to ignored files by @AAAkater in #12451
• Fix: The avatar and greeting message no longer appear in the Agent iFrame. [#12410] by @dcc123456 in #12459
• Refactor: import_test_cases use bulk_create by @Woody-Hu in #12456
• Docs: Fixed a display issue. by @writinwaters in #12463
• Fix: try handle authorization as api-token by @Lynn-Inf in #12462
• Fix: update uv python installation to version 3.12 in Dockerfile by @buua436 in #12464
• refactor: migrate env prefix to VITE_* by @JimmyBenKlieve in #12466
• fix: task executor with status "timeout" corrupts page when checking its details by @JimmyBenKlieve in #12467
• Feat: adapt to , arglist by @Lynn-Inf in #12468
• docs: update docs icons by @JimmyBenKlieve in #12465
• Fix: Issues and style fixes related to the 'Memory' page by @dcc123456 in #12469
• fix: prevent MinIO health check failure in multi-bucket mode by @maoxiaowang in #12446
• Fix: display agent name for extract messages by @Lynn-Inf in #12480
• Feat: Memory-message supports categorized display by @dcc123456 in #12482
• Fix: Fixed an issue where ESLint suggestions were not working in VS Code after upgrading to Vite. #12483 by @cike8899 in #12484
• Feat: support context window for docx by @Magicbook1108 in #12455
• Fix: toc async issue. by @KevinHuSh in #12485
• Feat: The chat feature supports streaming output, displaying results one by one. #12490 by @cike8899 in #12493
• Fix:Bedrock assume_role auth mode fails with LiteLLM "Extra inputs are not permitted" error by @Woody-Hu in #12495
• Doc: memory http api by @Lynn-Inf in #12499
• optimize doc id check: do not query db when doc id to validate is empty by @paulluap in #12500
• Refactor: Enhance delta streaming in chat functions for improved reasoning and content handling by @buua436 in #12453
• Refa(sdk/python/test): remove unused testcases and utilities by @asiroliu in #12505
• Fix: apply kb setting while re-parsing.... by @KevinHuSh in #12501
• Fix: add multimodel models in chat api by @Magicbook1108 in #12496
• Fix adding column error by @JinHai-CN in #12503
• Refactor: Replace Ant Design with shadcn in SparkModal, TencentCloudModal, HunyuanModal, and GoogleModal. #1036 by @cike8899 in #12510
• docs: fix embedding model switching tooltip by @lys1313013 in #12517
• Fix: "AttributeError("'list' object has no attribute 'get'")" by @Woody-Hu in #12518
• Fix: Bugs fixed by @dcc123456 in #12524
• Refactor: Refactoring OllamaModal using shadcn. #1036 by @cike8899 in #12530
• fix: keep password in opendal config to fix connection initialization by @whhe in #12529
• Fix: Baiduyiyan key invaild by @dcc123456 in #12531
• Feat: The transla...

v0.23.1

v0.23.1

Released on December 31, 2025.

Improvements

• Memory: Enhances the stability of memory extraction when all memory types are selected.
• RAG: Refines the context window extraction strategy for images and tables.

Fixed issues

• Memory:
  • The RAGFlow server failed to start if an empty memory object existed.
  • Unable to delete a newly created empty Memory.
• RAG: MDX file parsing was not supported.

Data sources

• GitHub
• Gitlab
• Asana
• IMAP

What's Changed

• Fix error in docs by @JinHai-CN in #12269
• docs: add security warnings for default passwords in .env by @RinCodeForge927 in #12250
• Update deploy_local_llm.mdx vllm guide picture by @yiminghub2024 in #12275
• Update deploy_local_llm.mdx by @yingfeng in #12276
• Fix: Data-source S3 page style by @dcc123456 in #12255
• Fix IDE warnings by @JinHai-CN in #12281
• Fix: init memory size from es by @Lynn-Inf in #12282
• Refactor: keep timestamp consistency by @Woody-Hu in #12279
• Fix the consistency of ts and datetime by @JinHai-CN in #12288
• Fix: Delete chunk images on document parser config change. by @lys1313013 in #12262
• feat: add MDX file support by @lys1313013 in #12261
• Fix: If there is an error message on the chat page, the subsequent message references will not display correctly. #12252 by @cike8899 in #12283
• Fix (sdk): ensure variables defined in rm_chunk API by @maoxiaowang in #12274
• security: replace unsafe eval with ast.literal_eval in vision operators by @RinCodeForge927 in #12236
• Feat: Add Asana data source integration and configuration options by @buua436 in #12239
• helm: improvements by @LyleLaii in #10976
• refactor: docling parser will close bytes io by @Woody-Hu in #12280
• Feat: github connector by @Magicbook1108 in #12292
• Feat: Gitlab connector by @Magicbook1108 in #12248
• Revert "Feat: github connector" by @KevinHuSh in #12296
• Fix: pick message to delete by @Lynn-Inf in #12295
• Fix: S3 parameter error by @dcc123456 in #12290
• docs: add optional proxy arguments for Docker build instructions by @maoxiaowang in #12272
• Feat: Remove the code that outputs jsonschema from the webhook.#10427 by @cike8899 in #12297
• docs: fix typos by @lys1313013 in #12301
• Refactor code by @JinHai-CN in #12305
• api key delete bug - Bug #3045 by @raminmardani in #12299
• Fix: use async task to save memory by @Lynn-Inf in #12308
• Fix TypeError in meta_filter when using numeric metadata by @tasumu in #12286
• Fix: Dataset parse error by @dcc123456 in #12310
• Fix IDE warnings by @JinHai-CN in #12315
• Feat: GitHub connector by @Magicbook1108 in #12314
• Fix: judge index exist before delete memory message. by @Lynn-Inf in #12318
• Feat: On the agent page and chat page, you can only select knowledge bases that use the same embedding model. #12320 by @cike8899 in #12321
• Feat: add IMAP data source integration with configuration and sync capabilities by @buua436 in #12316
• Update release workflow by @JinHai-CN in #12327
• Update model providers by @JinHai-CN in #12333
• Potential fix for code scanning alert no. 62: Workflow does not contain permissions by @yingfeng in #12334
• Update release workflow by @JinHai-CN in #12335
• fix: malformed dynamic translation key chunk.docType.${chunkType} by @JimmyBenKlieve in #12329
• Feat: Adapt the theme of the documentation page. #10427 by @cike8899 in #12337
• Fix: Dataset parse logic by @dcc123456 in #12330
• Feat: Display mode at the begin node #10427 by @cike8899 in #12326
• Refine: image/table context. by @KevinHuSh in #12336
• Docs: What is RAG & What is Agent context engine by @writinwaters in #12341
• fix(admin-ui): pagination auto reset to first page when after refetching data by @JimmyBenKlieve in #12339
• fix: correct session reference initialization to prevent dialogue misalignment by @DanielDerefaka in #12343
• Fix: Bug fixed by @dcc123456 in #12345
• Feat: Fixed the issue where the newly created agent begin node displayed "undefined". #10427 by @cike8899 in #12348
• [Feature] Admin: sort user list by email by @JinHai-CN in #12350
• Fix: apply kb configured llm issue. by @KevinHuSh in #12354
• [Bug] Don't display not used component status in admin status dashboard by @JinHai-CN in #12355
• Update release notes by @JinHai-CN in #12356
• Docs: Update version references to v0.23.1 in READMEs and docs by @asiroliu in #12349
• Remove doc of health check by @JinHai-CN in #12363
• fix(rag/prompts): Restructure metadata extraction rules for precision by @asiroliu in #12360
• Fix: Batch parsing problem by @dcc123456 in #12358
• Add docs category file by @JinHai-CN in #12359

New Contributors

• @RinCodeForge927 made their first contribution in #12250
• @lys1313013 made their first contribution in #12262
• @maoxiaowang made their first contribution in #12274
• @LyleLaii made their first contribution in #10976
• @raminmardani made their first contribution in #12299
• @tasumu made their first contribution in #12286
• @DanielDerefaka made their first contribution in #12343

Full Changelog: v0.23.0...v0.23.1

v0.23.0

v0.23.0

Released on December 27, 2025.

New features

• Memory
  • Implements a Memory interface for managing memory.
  • Supports configuring context via the Retrieval or Message component.
• Agent
  • Improves the Agent component's performance by refactoring the underlying architecture.
  • The Agent component can now output structured data for use in downstream components.
  • Supports using webhook to trigger agent execution.
  • Supports voice input/output.
  • Supports configuring multiple Retrieval components per Agent component.
• Ingestion pipeline
  • Supports extracting table of contents in the Transformer component to improve long-context RAG performance.
• Dataset
  • Supports configuring context window for images and tables.
  • Introduces parent-child chunking strategy.
  • Supports auto-generation of metadata during file parsing.
• Chat: Supports voice input.

Improvements

• Bumps RAGFlow's document engine, [Infinity](GitHub - infiniflow/infinity: The AI-native database built for LLM applications, providing incredibl) to v0.6.15 (backward compatible).

Data sources

• Google Cloud Storage
• Gmail
• Dropbox
• WebDAV
• Airtable

Model support

• GPT-5.2
• GPT-5.2 Pro
• GPT-5.1
• GPT-5.1 Instant
• Claude Opus 4.5
• MiniMax M2
• GLM-4.7.
• A MinerU configuration interface.
• AI Badgr (model provider).

API changes

HTTP API

• Converse with Agent returns complete execution trace logs.
• Create chat completion supports metadata-based filtering.
• Converse with chat assistant supports metadata-based filtering.

What's Changed

• fix cohere rerank base_url default by @HeyPhiS in #11353
• Feat: add OceanBase doc engine by @whhe in #11228
• fix(llm): handle None response in total_token_count_from_response by @zhangzhefang-github in #10941
• Feat: Add TCADP parser for PPTX and spreadsheet document types. by @aidansu in #11041
• Locale: update russian language by @rjohny55 in #11393
• locale en add russian language option by @rjohny55 in #11392
• Feature (canvas): Add mind tagging support by @huang-aoqin in #11359
• Use array syntax for commands in docker-compose-base.yml by @panzi in #11391
• Bump infinity to 0.6.6 by @yuzhichang in #11399
• Fix: Introducing a new JSON editor by @dcc123456 in #11401
• Fix:RagFlow not starting with Postgres DB by @buua436 in #11398
• Fix: Optimize edge check & incorrect parameter usage by @Magicbook1108 in #11396
• Feat: support union logic operations for meta data filters. by @KevinHuSh in #11404
• Fix: Change package-lock.json by @dcc123456 in #11407
• Fix: incorrect retrieval total count with pagination enabled by @yongtenglei in #11400
• Feat: Outputs data is directly synchronized to the canvas without going through the form. #10427 by @cike8899 in #11406
• Fix:improve multi-column document detection by @buua436 in #11415
• Feat: add more chunking method by @Magicbook1108 in #11413
• Fix: no result if metadata returns none. by @KevinHuSh in #11412
• Feat: enriches Notion connector by @yongtenglei in #11414
• Fix: missing parameters in by_plaintext method for PDF naive mode by @it-coder in #11408
• Fix: Modify the style of the user center #10703 by @dcc123456 in #11419
• component horizontal padding adjustment by @JimmyBenKlieve in #11418
• Feat: Add a loop variable to the loop operator. #10427 by @cike8899 in #11423
• Fix: UI adjustments, replacing private components with public components by @dcc123456 in #11438
• Fix: improve PDF text type detection by expanding regex content by @FallingSnowFlake in #11432
• Feat: ignore chunk size when using custom delimiters by @yongtenglei in #11434
• Feat: API supports toc_enhance. by @KevinHuSh in #11437
• Feat: Enable logical operators in metadata. #11387 #11376 by @cike8899 in #11442
• Fix:Resolves the issue of sessions not being saved when the variable is array. by @dcc123456 in #11446
• Feat: add auth header for Ollama chat model by @yongtenglei in #11452
• Add Moodle data source integration by @levischd in #11325
• Bump beartype to 0.22.6 by @yuzhichang in #11463
• Update logging for auto-generated SECRET_KEY by @FoolFu in #11458
• Fix nginx startup failure in HTTPS mode (host not found) by @davidlcarrascal in #11455
• Feat: Implement temporary conversation removal logic in ConversationD… by @Zwanan-github in #11454
• Fix:Modify the personal center style #10703 by @dcc123456 in #11470
• Fix: coroutine object has no attribute get by @Magicbook1108 in #11472
• Docs: Updated use_kg and toc_enhance switch descriptions by @writinwaters in #11485
• Allow create super user when start rag server. by @Ox0400 in #10634
• Feat: add datasource Dropbox by @yongtenglei in #11488
• Refactor: Email parser use with to handle buffer by @Woody-Hu in #11496
• Fix: After saving the model parameters of the chat page, the parameter disappears. #11500 by @cike8899 in #11501
• Fix: unexpected parameter. by @KevinHuSh in #11497
• Fix: Modify the style of your personal center #10703 by @dcc123456 in #11487
• Feat: support operator in/not in for metadata filter. by @KevinHuSh in #11503
• Feat: support operator in/not in for metadata filter. #11376 #11378 by @cike8899 in #11506
• Chore(deps): Bump pypdf from 6.0.0 to 6.4.0 by @dependabot[bot] in #11505
• Fix: code exec component vulnerability and add support for nested list and dict object by @yongtenglei in #11504
• Fix: uv lock updates by @KevinHuSh in #11511
• Feat: add addressing style config for S3-compatible storage by @jacy1101 in #11510
• Refactoring: Integrating the file preview component by @dcc123456 in #11523
• Fix: incorrect image merging for naive markdown parser by @yongtenglei in #11520
• Fix: excel default optimization. by @KevinHuSh in #11519
• Bump infinity to 0.6.7 by @yuzhichang in #11528
• Use Infinity single-field-multi-index by @yuzhichang in #11444
• Feat: Add Webdav storage as data source by @Jonah879 in #11422
• Fix: Refactoring and enhancing the functionality of the delete confirmation dialog component #10703 by @dcc123456 in #11542
• feat: improve metadata handling in connector service by @levischd in #11421
• Fix issue where HTML file parsing may lose content. by @myoldcat in #11536
• fix(files): replace hard coded status codes with constants by @huang-aoqin in #11544
• Minor style changes by @writinwaters in #11554
• Feat: add context for figure and table by @yongtenglei in #11547
• Fix ft_title_rag_fine by @yuzhichang in #11555
• Feat:new component Loop by @buua436 in #11447
• Feat: Add loop operator node. #10427 by @cike8899 in #11449
• Fix: enable structured output for agent with tool by @buua436 in #11558
• Feat: Delete useless knowledge base, chat, and search files. #10...

v0.22.1

Release v0.22.1 created from cfdcceb at 2025-11-19 20:13:52+08:00

Summary

Improvements

• Agent:
  • Supports exporting Agent outputs in Word or Markdown formats.
  • Adds a List operations component.
  • Adds a Variable aggregator component.
• Data sources:
  • Supports S3-compatible data sources, e.g., MinIO.
  • Adds data synchronization with JIRA.
• Continues the redesign of the Profile page layouts.
• Upgrades the Flask web framework from synchronous to asynchronous, increasing concurrency and preventing blocking issues caused when requesting upstream LLM services.

Fixed issues

• A v0.22.0 issue: Users failed to parse uploaded files or switch embedding model in a dataset containing parsed files using a built-in model from a -full RAGFlow edition.
• Image concatenated in Word documents. #11310
• Mixed images and text were not correctly displayed in the chat history.

Newly supported models

• Gemini 3 Pro Preview

What's Changed

• Refactor function name by @JinHai-CN in #11210
• Fix:reset the agent component’s output by @buua436 in #11222
• Fix: Profile picture cropping supported #10703 by @dcc123456 in #11221
• Fix: update readme by @Magicbook1108 in #11212
• Feature/doc upload api add parent path 20251112 by @yngvarhuang in #11231
• Refa: rm useless code. by @KevinHuSh in #11238
• Feat:support API for generating knowledge graph and raptor by @buua436 in #11229
• Test: handle duplicate names by appending "(1)" by @asiroliu in #11244
• Fix: Law parser by @Magicbook1108 in #11246
• Minor tweaks by @JinHai-CN in #11249
• Feat: add fault-tolerant mechanism to RAPTOR by @yongtenglei in #11206
• Feat:update check_embedding api by @buua436 in #11254
• Docs: default admin credentials by @writinwaters in #11260
• Refactor service config fetching in admin server by @JinHai-CN in #11267
• Fixes: Added session variable types and modified configuration by @dcc123456 in #11269
• Remove leftover account and password from the code by @redredrrred in #11248
• Fix: rerank algorithm by @Magicbook1108 in #11266
• Fix: check task executor alive and display status by @Lynn-Inf in #11270
• Refactor: improve mineru_parser get property logic by @Woody-Hu in #11268
• Feat:new component list operations by @buua436 in #11276
• Fix: Added the ability to download files in the agent message reply function. by @dcc123456 in #11281
• Fix:ListOperations does not support sorting arrays of objects. by @buua436 in #11278
• Feat: extract message output to file by @Magicbook1108 in #11251
• Doc: add default username & pwd by @Magicbook1108 in #11283
• Remove unused code and fix performance issue by @JinHai-CN in #11284
• Minor tweats by @JinHai-CN in #11271
• Feat: add Jira connector by @yongtenglei in #11285
• Doc: style fix by @Magicbook1108 in #11295
• Fix: Fixed the issue of not being able to select the time zone in the user center. by @dcc123456 in #11298
• Fix: default model base url extraction logic by @sd109 in #11263
• Fix: create dataset return type inconsistent by @Magicbook1108 in #11272
• Refactor: move mcp connection utilities to common by @JinHai-CN in #11304
• Fix: concat images in word document. by @Magicbook1108 in #11310
• Feat: Construct a dynamic variable assignment form #10427 by @cike8899 in #11316
• feat: add new LLM provider Jiekou.AI by @cnJasonZ in #11300
• Fix: Fixed an issue where adding session variables multiple times would overwrite them. by @dcc123456 in #11308
• Feat: add s3-compatible storage boxes by @Jonah879 in #11313
• fix: update check_embedding failed info by @buua436 in #11321
• fix: entrypoint.sh typo for disable datasync command by @0xlucasliao in #11326
• Feat: update version to 0.22.1 by @Lynn-Inf in #11331
• Feat: Fixed the issue where form data assigned by variables was not updated in real time. #10427 by @cike8899 in #11333
• Fix: add describe_image_with_prompt for ZHIPU AI by @Magicbook1108 in #11317
• Fix: manual parser with mineru by @Magicbook1108 in #11336
• Fix:limit random sampling range in check_embedding by @buua436 in #11337
• Fix: No results can be found through the API /api/v1/dify/retrieval by @yongtenglei in #11338
• Fix: crop index may out of range by @yongtenglei in #11341
• Feat: Alter flask to Quart for async API serving. by @KevinHuSh in #11275
• Feat: Add a switch to control the display of structured output to the agent form. #10427 by @cike8899 in #11344
• Feat: new component variable assigner by @buua436 in #11050
• Feat: Display variables in the variable assignment node. #10427 by @cike8899 in #11349
• Fix: Modify the personal center style #10703 by @dcc123456 in #11347
• Fix: variable assigner issue. by @KevinHuSh in #11351
• Update version info to v0.22.1 by @JinHai-CN in #11346
• Fix: Fixed an issue where variable aggregation operators could not be connected to other operators. #10427 by @cike8899 in #11358
• Refactor /stats API by @JinHai-CN in #11363
• chore(template): introducing variable aggregator to customer service template by @ZhenhangTung in #11352
• Feat: add Gemini 3 Pro preview by @yongtenglei in #11361
• Fix: component list operation issue. by @KevinHuSh in #11364
• Fix: bbox not included in mineru output by @Magicbook1108 in #11365
• Feat: Set the outputs type of list operation. #10427 by @cike8899 in #11366
• Update README for supporting Gemini 3 Pro by @JinHai-CN in #11369
• Add release notes by @JinHai-CN in #11372
• Feat: Structured data will still be stored in outputs for compatibility with older versions. #10427 by @cike8899 in #11368
• Fix: circle imports issue. by @KevinHuSh in #11374
• Feat: The key for the begin operator can only contain alphanumeric characters and underscores. #10427 by @cike8899 in #11377
• Doc: Added v0.22.1 release notes by @writinwaters in #11383
• Fix: refine error msg. by @KevinHuSh in #11380
• Feat: If a query variable in a data manipulation operator is deleted, a warning message should be displayed to the user. #10427 #11255 by @cike8899 in #11384
• Fix some multilingual issues by @dcc123456 in #11382
• Doc: Optimize read me by @Magicbook1108 in #11386
• Docs: minor by @writinwaters in #11385
• Feat: Fixed an issue where modifying fields in the agent operator caused the loss of structured data. #10427 by @cike8899 in #11388

New Contributors

• @cnJasonZ made their first contribution in #11300
• @0xlucasliao made their first contribution in #11326

Full Changelog: v0.22.0...v0.22.1

v0.22.0

Released on November 12, 2025.

Breaking Changes

• From this release onwards, we ship only the slim edition (without embedding models) Docker image and no longer append the -slim suffix to the image tag.

New Features

• Dataset:
  • Supports data synchronization from five online sources (AWS S3, Google Drive, Notion, Confluence, and Discord).
  • RAPTOR can be built across an entire dataset or on individual documents.
• Ingestion pipeline: Supports Docling document parsing in the Parser component.
• Launches a new administrative Web UI dashboard for graphical user management and service status monitoring.
• Agent:
  • Supports structured output.
  • Supports metadata filtering in the Retrieval component.
  • Introduces a Variable aggregator component with data operation and session variable definition capabilities.

Improvements

• Agent: Supports visualizing previous components' outputs in the Await Response component.
• Revamps the model provider page.
• Upgrades RAGFlow's document engine Infinity to v0.6.5.

Added Models

• Kimi-K2-Thinking

New agent templates

• Interactive Agent, incorporates real-time user feedback to dynamically optimize Agent output.

What's Changed

• Feat: add Docling parser by @buua436 in #10759
• Move test files by @JinHai-CN in #10765
• Don't release full image by @yuzhichang in #10654
• Fix:ERROR 20 Method rag.nlp.search.Dealer.search() parameter highlight="None" violates type hint bool | list, as <class "builtins.NoneType"> "None" not list or bool. by @Woody-Hu in #10743
• Fix: prio synonym match than wordnet for english by @Magicbook1108 in #10762
• Fix: Fixed the issue where dataset log avatars were displayed incorrectly #9869 by @cike8899 in #10764
• Docs/supplement incomplete params by @jiasu-hezhip in #10758
• Fix: Opendal miss tenant id by @Woody-Hu in #10774
• Feat: Adjust the style of the canvas node #10703 by @cike8899 in #10795
• Removing value={field} from EditTag by @adrianad in #10767
• Fixed the bug that passing an empty array will not update by @jiasu-hezhip in #10755
• More doc on RAGFlow image by @yuzhichang in #10800
• Fix: gemini cv model chat issue. by @KevinHuSh in #10799
• Doc: readme updates. by @KevinHuSh in #10801
• Fix:Error retrieving DOCX image (docx.image.exceptions.UnrecognizedImageError) by @Woody-Hu in #10794
• Feat: add TCADP Parser by @aidansu in #10775
• Fix: Home and team page style adjustment, and some bug fixes #10703 by @dcc123456 in #10805
• Fix: Pass kwargs in python api #10699 by @Magicbook1108 in #10808
• Feat: MinerU supports VLM-Transfomers backend by @yongtenglei in #10809
• Feat: Adjust the style of the toolbar at the bottom of the agent canvas #10703 by @cike8899 in #10807
• Feat:All-in-one MinerU and Docling by @buua436 in #10813
• Fix: bug fixes and icon replacement #10703 by @dcc123456 in #10814
• Feat: Modify the style of the agent operator form #10703 by @cike8899 in #10821
• Doc: Updated How to update MinerU's settings by @writinwaters in #10818
• Fix: parsing excel with chartsheet & Clamp begin to a minimum of 0 to prevent negative indexing by @Magicbook1108 in #10819
• Refactor: rename rmSpace to remove_redundant_spaces by @JinHai-CN in #10796
• Feature/add new models for token pony and bug fix for use llm by @zhileihuang in #10823
• Fix: Add video icon in knowledge base #10703 by @dcc123456 in #10832
• Fix:remove unexpected keyword argument in table_structure_recognizer logging by @buua436 in #10831
• Customize service ports in tests.yml by @yuzhichang in #10834
• Fix ragflow image by @yuzhichang in #10838
• Fix: logging issues by @Magicbook1108 in #10836
• Add time utils by @JinHai-CN in #10849
• Feat: Adjust the style of the agent operator form tool #10703 by @cike8899 in #10841
• Fix: Refactor the similarity slider component and modify the style of the dataset-test page #10703 by @dcc123456 in #10846
• Add readme in web by @JinHai-CN in #10855
• Remove 'DID YOU KNOW', when start front-end by @JinHai-CN in #10853
• Feat: Admin UI by @JimmyBenKlieve in #10857
• Update Octoverse to README by @yingfeng in #10859
• Refactor: Improve how to get total token count for AnthropicCV by @Woody-Hu in #10658
• Test: update test cases for chunk retrieval pagination by @asiroliu in #10839
• Fix: Chat session completion by @Magicbook1108 in #10851
• Feat: allow initialize Redis without password by @yongtenglei in #10856
• Update admin client default port to 9381 by @JinHai-CN in #10862
• Update web/README.md by @JinHai-CN in #10864
• Feat: Configure structured data output for agent forms #10866 by @cike8899 in #10867
• Feat: rename file to avoid package name conflict by @Magicbook1108 in #10863
• Chore(config): remove Youdao and BAAI embedding model providers by @asiroliu in #10873
• Fix: predictable token generation by @yongtenglei in #10868
• Chore(docker): add standard HTTP/HTTPS and MCP ports to .env configuration by @asiroliu in #10881
• Feat: Allow other operators to reference the structured output defined by the agent operator. #10866 by @cike8899 in #10886
• Feat: add German translations for all agent templates and optimized line breaks for template titles by @Jonah879 in #10643
• Bump infinity to 0.6.2 by @yuzhichang in #10887
• Fix: opensearch retrieval error by @Magicbook1108 in #10891
• Mineru api support by @yoyokko in #10874
• Feat: location rule for admin UI by @Lynn-Inf in #10894
• Fix:enhance delimiters in markdown parser by @buua436 in #10896
• Fix: video parser should follow selected VLM in pipeline by @yongtenglei in #10900
• Fix: law parser by @Magicbook1108 in #10897
• Fix: the input length exceeds the context length by @Magicbook1108 in #10895
• Feat:check embedding model api by @buua436 in #10854
• Feat: The query variables of the subsequent operators can reference the structured variables defined in the agent operator. #10866 by @cike8899 in #10902
• Fix several admin UI issues by @JimmyBenKlieve in #10869
• Fix(ci): Add error handling to Docker image removal in tests workflow by @asiroliu in #10904
• Feat: Modify the style of the query variable dropdown list. #10866 by @cike8899 in #10903
• Refactor(setting-model): Refactor the model management interface and optimize the component structure. #10703 by @dcc123456 in #10905
• Refactor:improve the logic for rerank models to cal the total token count by @Woody-Hu in #10882
• Fix:KeyError: 'table_body' of mineru parser by @Woody-Hu in #10773
• Feat: Rename the files in the jsonjoy-builder directory to lowercase. #10866 by @cike8899 in #10908
• Fix: Fixed the styling and logic issues on the model provider page #10703 by @dcc123456 in #10909
• Fix: The nodes on the canvas were not updated in time after the operator name was modified. #10866 by @cike8899 in #10911
• Remove 'get_lan_ip' and add common misc_utils.py by @JinHai-CN in https://github.com/...

v0.21.1

Release v0.21.1 created from de24e74 at 2025-10-23 19:08:09+08:00

New features

• Experimental: Adds support for PDF document parsing using MinerU. See here.

Improvements

• Enhances UI/UX for the dataset and personal center pages.
• Upgrades RAGFlow's document engine, Infinity, to v0.6.1.

Fixed issues

• An issue with video parsing.

What's Changed

• Fix problem with Google Cloud models with reasoning (like gemini) - Additional fix to issue #10474 by @glu000 in #10502
• Fix: support auto width when print table by @Lynn-Inf in #10575
• Doc: update released tag. by @Lynn-Inf in #10578
• fix(_handle_entity_relation_summary): correctly calculate the descriptions_list by @ffffffff0 in #10534
• Fix (i18n): Update the Chinese and English description of RAPTOR functionality #9869 by @dcc123456 in #10581
• Refactor admin client for message prompts by @JinHai-CN in #10583
• Docs: Updated dataset configuration, KG building and RAPTOR building for v0.21.0 by @writinwaters in #10584
• Don't rerank for infinity by @yuzhichang in #10579
• Doc: Added Long context RAG guide by @writinwaters in #10591
• Use relative path to import same module by @JinHai-CN in #10587
• Improve file management by @yongtenglei in #10577
• Update admin client README and doc by @JinHai-CN in #10594
• Feat: Support Specifying OpenRouter Model Provider by @buua436 in #10550
• Build(deps): Bump axios from 1.7.2 to 1.12.0 in /web by @dependabot[bot] in #10393
• Fix (dataset setting): Remove the introduction and use of TagItems in the configuration. #9869 by @dcc123456 in #10595
• Fix: add toc_kwd field and update page_num_int type by @asiroliu in #10596
• Update architecture image and ragflow-cli version by @JinHai-CN in #10605
• Feat: add forgot password reset, solve #8547 by @Magicbook1108 in #10586
• Fix: Add RetryingPooledPostgresqlDatabase to handle max_retries param by @arogan178 in #10524
• Feat: Collapse the excess portion of the tool node and retrieval node #9869 by @cike8899 in #10604
• Fix: Bug fixes #9869 by @dcc123456 in #10600
• Feat: The bottom anchor of the agent node is only displayed when there is a downstream node #9869 by @cike8899 in #10611
• Feat: add forgot password reset (update naming style), solve #8547 by @Magicbook1108 in #10606
• Fix: Open the parser operator configuration, save it, and run the agent. An error will be reported. #10615 by @cike8899 in #10619
• Fix: search highlight. by @KevinHuSh in #10616
• Doc: minor by @writinwaters in #10627
• Feat: add MinerU parser by @yongtenglei in #10621
• Fix：display agents by @Lynn-Inf in #10620
• feat(storybook): Storybook with Calendar and Modal components #9869 by @dcc123456 in #10626
• Docs: minor by @writinwaters in #10630
• Feat:VolcEngine Model type add IMAGE2TEXT by @buua436 in #10629
• Feat: Delete useless files from the data pipeline #9869 by @cike8899 in #10632
• Fix: Restore the sidebar description of DP slicing method #9869 by @dcc123456 in #10633
• File: Now parsing support all types of embedded documents, solved #10059 by @Magicbook1108 in #10635
• Doc: miscellaneous by @writinwaters in #10641
• Feat: Display the pipeline on the agent canvas #9869 by @cike8899 in #10638
• Fix：Text color in Floating Widget (Intercom-style) #10624 by @dcc123456 in #10639
• Feat：admin api by @Lynn-Inf in #10642
• feat: implement CLI of role-based access control system by @JinHai-CN in #10650
• Supports running single command by @JinHai-CN in #10651
• Fix: retrieval test by @KevinHuSh in #10662
• Fix: Remove pdf embed support, update based on #10635 by @Magicbook1108 in #10663
• Fix: The agent dialogue sheet does not display the opening remarks. #10664 by @cike8899 in #10665
• Fix: Optimize code and fix ts type errors #9869 by @dcc123456 in #10666
• Feat: Gemini supports video parsing by @yongtenglei in #10671
• Feat: Qwen-VL series supports video parsing by @yongtenglei in #10676
• Feat:Vision Model Image Enhancement in Manual/Paper/Book/One chunker by @buua436 in #10640
• Update message printout when start ingestion server by @JinHai-CN in #10677
• Fix(edit-tag): Fix the bug that the edit-tag tag cannot be deleted #9869 by @dcc123456 in #10679
• Feat:setting metadata in the retrieval by @buua436 in #10682
• Fix: Excel2HTML can't support XLS（Excel 97-2003） by @Woody-Hu in #10660
• Feat: Support attribute filtering #8703 by @Magicbook1108 in #10670
• Feat: Make knowledge base renaming automatically reflected in agent discussions, solved #10597 by @Magicbook1108 in #10680
• Feat: Display the pipeline operation sheet on the agent page #9869 by @cike8899 in #10690
• Fix: tree merge by @Magicbook1108 in #10691
• Test: update test cases for chunk retrieval pagination by @asiroliu in #10694
• Feat: A pipeline's child node can only have one node #9869 by @cike8899 in #10695
• Feat: Move the pipeline translation field to flow #9869 by @cike8899 in #10697
• Fix: potential negative max_tokens in RAPTOR by @yongtenglei in #10701
• Feat: Change the style of all cards according to the design #10703 by @cike8899 in #10704
• Fix:wrong param in manual chunk by @buua436 in #10710
• Fix: Profile page UI adjustment #9869 by @dcc123456 in #10706
• Docs: Added token chunker and title chunker components by @writinwaters in #10711
• Feat: add fault-tolerant mechanism to GraphRAG by @yongtenglei in #10708
• Refa: improve flow of GraphRAG and RAPTOR by @yongtenglei in #10709
• fix:RAGFlowOSS.put() got an unexpected keyword argument 'tenant_id' by @Woody-Hu in #10712
• Feat: Adjust the style of the mcp dialog #10703 by @cike8899 in #10719
• Fix: can't upload image in ollama model #10447 by @Magicbook1108 in #10717
• Refactor: How LiteLLMBase Calculate total count by @Woody-Hu in #10532
• hotfix: Rename chunk summary's component name by @ZhenhangTung in #10721
• Fix: Optimize the style of the personal center sidebar component #9869 by @dcc123456 in #10723
• bump infinity to v0.6.1 by @yuzhichang in #10727
• Feat: Display the video field in the parser operator #9869 by @cike8899 in #10728
• Feat: version 0.21.1 by @Lynn-Inf in #10718
• Fix: model parameter by @Magicbook1108 in #10730
• Fix：filename is not displayed on the overview page #9869 by @dcc123456 in #10731
• Fix: add video parser by @KevinHuSh in #10735
• Feat: pipeline supports MinerU PDF parser by @yongtenglei in #10736
• Fix typo by @adrianad in #10737
• Fix: video file suffix by @KevinHuSh in #10740
• Feat: Add suffi...

v0.21.0

Released on October 15, 2025.

New features

• Orchestratable ingestion pipeline: Supports customized data ingestion and cleansing workflows, enabling users to flexibly design their data flows or directly apply the official data flow templates on the canvas.
• GraphRAG & RAPTOR write process optimized: Replaces the automatic incremental build process with manual batch building, significantly reducing construction overhead.
• Long-context RAG: Automatically generates document-level table of contents (TOC) structures to mitigate context loss caused by inaccurate or excessive chunking, substantially improving retrieval quality. This feature is now available via a TOC extraction template.
• Admin CLI: Introduces a new command-line tool for system administration, allowing users to manage and monitor RAGFlow's service status via command line.

Improvements

• Redesigns RAGFlow's Login and Registration pages.
• Upgrades RAGFlow's document engine Infinity to v0.6.0.

Added models

• Tongyi Qwen 3 series
• Claude Sonnet 4.5
• Meituan LongCat-Flash-Thinking

New agent templates

• Company Research Report Deep Dive Agent: Designed for financial institutions to help analysts quickly organize information, generate research reports, and make investment decisions.
• Orchestratable Ingestion Pipeline Template: Allows users to apply this template on the canvas to rapidly establish standardized data ingestion and cleansing processes.

What's Changed

• Feat: dataflow supports Spreadsheet and Word processor document by @yongtenglei in #9996
• Feat: dataflow supports markdown by @Lynn-Inf in #10003
• Refactor:Improve BytesIO usage for image2base64 by @Woody-Hu in #9997
• Fix: Fixed the issue where newly added tool operators would disappear after editing the form #10013 by @cike8899 in #10016
• Feat: Add type card to create agent dialog #9869 by @cike8899 in #10025
• Feat(config): Update service config template new defaults by @asiroliu in #10029
• Feat: Import dsl from agent list page #9869 by @cike8899 in #10033
• feat: Added UI functions related to data-flow knowledge base #3221 by @dcc123456 in #10038
• Refactor:Improve BytesIO usage for GeminiCV by @Woody-Hu in #10042
• Feat: Translate the parser operator #9869 by @cike8899 in #10037
• Fix: In ragflow/rag/app /naive.py, if there are multiple images in one line, the other images will be lost by @Woody-Hu in #9968
• Fix: token num exceed by @Lynn-Inf in #10046
• Fix: Hide dataflow related functions #9869 by @dcc123456 in #10045
• The chat dialog box cannot be fully displayed on a small screen #10034 by @cike8899 in #10049
• Feat: add TokenPony model provider by @zhileihuang in #9932
• Docs: How to connect to an MCP server as a client by @writinwaters in #10043
• Feat: Agent component support inserting variables(#10048) by @wilmerxu in #10055
• Feat: dataflow supports text by @Lynn-Inf in #10058
• Feat: Translate the fields of the embedded dialog box on the agent page #3221 by @cike8899 in #10072
• Fix: terminate onnx inference session manually by @Lynn-Inf in #10076
• feat: add multiple docker tags (latest, latest_full, latest_slim) to … by @adrianweidig in #10040
• Fix: The same model appears twice in the drop-down box. #10102 by @cike8899 in #10103
• Fix: judge not empty before delete by @Lynn-Inf in #10099
• Feat: add support for the Ascend layout recognizer by @yongtenglei in #10105
• Docs: Added an Execute SQL tool reference by @writinwaters in #10108
• Fix(agent template): wrap template variables in curly braces by @asiroliu in #10109
• Feat: add support for the Ascend table structure recognizer by @yongtenglei in #10110
• Feat/parse img by @Lynn-Inf in #10112
• Fix: set default chunk_token_num in html_parser by @buua436 in #10118
• Fix: usage of postgresql -> postgres for db_type by @cucusenok in #10120
• Docs: Removed /v1 from Ollama base URLs by @writinwaters in #10067
• Fixes session_id passing in agent_openai completion. by @bluthen in #10124
• Feat: parse audio by @Lynn-Inf in #10133
• Build(deps): Bump axios from 1.9.0 to 1.12.0 in /sandbox/sandbox_base_image/nodejs by @dependabot[bot] in #10091
• Support image recognition for url links in Markdown file, fix log error in code_exec by @Magicbook1108 in #10139
• Correct the text of vectorSimilarityWeight in zh.ts by @aopstudio in #10128
• Docs: Updated database types supported by the Execute SQL tool by @writinwaters in #10113
• Feat: add CometAPI to LLMFactory and update related mappings by @tensornull in #10119
• Feat: add support for KB document basic info by @yongtenglei in #10134
• Fix: embedding model failure in CometAPI by @buua436 in #10137
• Fix: handle zero by @Lynn-Inf in #10149
• Fix: Merge different types of models from the same manufacturer #10146 by @cike8899 in #10157
• Feat: migrate OpenAI-compatible chats to LiteLLM by @yongtenglei in #10148
• Fix: knowledge base's embedded model form layout and dependency imports in the main branch. #9869 by @dcc123456 in #10160
• Add Firecrawl integration for RAGFlow by @aybanda in #10152
• Feat: Add file convert to document API just like file2document_app.py by @huang-aoqin in #10158
• Support server health check by http://localhost:/v1/system/healthz by @Magicbook1108 in #10150
• Docs: Input and output variables defined in the Input and Output sections must also be implemented in your code. by @writinwaters in #10162
• fix: correct rerank_model condition logic by @buua436 in #10174
• Docs: How to accelerate question answering by @writinwaters in #10179
• Feat: add support for Anthropic third-party API by @buua436 in #10173
• perf: add get_all_kb_doc_count func to simplify kb.doc_num updating by @whhe in #10169
• Fix: skip tag query if tag kbs are invalid by @whhe in #10168
• Fix: enable scrolling at chat setting by @atsushi729 in #10184
• Feat: add admin CLI and admin service by @JinHai-CN in #10186
• Chat Widget by @melmathari in #10187
• Add tree_merge for law parsers, significantly outperforming hierarchical_merge by @Magicbook1108 in #10202
• Refactor: use the same implement for total token count from res by @Woody-Hu in #10197
• fix python_api example by @zhangwfjh in #10196
• The retrieval component needs to support returning JSON data(#10170) by @wilmerxu in #10171
• Fix: resolve hash collisions by switching to UUID &correct logic for always-true statements & Update GPT api integration & Support qianwen-deepresearch by @Magicbook1108 in #10208
• Fix: string merge bug about agent TextProcessing. (\m) by @zhimeng123 in #10211
• Fix bug: broken import from rag.prompts.prompts by @JinHai-CN in #10217
• Fix: azure OpenAI retry by @SgtMarmite in #10213
• revert gpt5 integration by @Magicbook1108 in #10228
• Flow text processing bug by @zhimeng123 in #10246
• fix: prevent list index out of range in chat streaming by @buua436 in #10238
• UI: Add Russian language by @rjohny55 in #10249
• Fix:max_tokens must be at least 1, got -950, BadRequestError by @Woody-Hu in #10252
• UI: Update Russian lang...

v0.20.5

Summary

Improvements

• Agent Performance Optimized: Improved planning and reflection speed for simple tasks; optimized concurrent tool calls for parallelizable scenarios, significantly reducing overall response time.
• Agent Prompt Framework exposed: Developers can now customize and override framework-level prompts in the system prompt section, enhancing flexibility and control.
• Execute SQL Component Enhanced: Replaced the original variable reference component with a text input field, allowing free-form SQL writing with variable support.
• Chat: Re-enabled Reasoning and Cross-language search.
• Retrieval API Enhanced: Added metadata filtering support to the Retrieve chunks method.

Added models

• Meituan LongCat
• Kimi: kimi-k2-turbo-preview and kimi-k2-0905-preview
• Qwen: qwen3-max-preview
• SiliconFlow: DeepSeek V3.1

Fixed issues

• Dataset: Deleted files remained searchable.
• Chat: Unable to chat with an Ollama model.
• Agent: Resolved issues including cite toggle failure, task mode requiring dialogue triggers, repeated answers in multi-turn dialogues, and duplicate summarization of parallel execution results.

What's Changed

• Fix typo by @JinHai-CN in #9766
• Feature/workflow en cn by @YyBoom233 in #9742
• Fix: second round value removal. by @KevinHuSh in #9756
• Refactor: Improve the try logic for upload_to_minio by @Woody-Hu in #9735
• add Russian in translation table index.tsx by @rjohny55 in #9773
• Clarify installation of pre-commit alongside uv in README by @aopstudio in #9749
• Feat: Add AvatarUpload component #3221 by @cike8899 in #9777
• Add Russian language Update app.tsx by @rjohny55 in #9772
• The agent directly outputs the results under the task model #9745 by @cike8899 in #9746
• Improve model tag rendering by splitting comma-separated string into styled components by @aopstudio in #9762
• Fix: agent template error. by @KevinHuSh in #9784
• Feat: Use AvatarUpload to replace the avatar settings on the dataset and search pages #3221 by @cike8899 in #9785
• Fix: Optimized variable node display and Agent template multi-language support #3221 by @dcc123456 in #9787
• Refa: unify reference format of agent completion and OpenAI-compatible completion API by @yongtenglei in #9792
• Fix: Fixed the issue where the thinking mode on the chat page could not be turned off #9789 by @cike8899 in #9794
• Feat: init dataflow. by @KevinHuSh in #9791
• Feat: Extract the save buttons for dataset and chat configurations to separate files to increase permission control #3221 by @cike8899 in #9803
• Fix: Optimized Input and MultiSelect component functionality and dataSet-chunk page styling #9779 by @dcc123456 in #9815
• Fix: Ollama chat cannot access remote deployment by @yongtenglei in #9816
• Fix: Fixed the issue that similarity threshold modification in chat and search configuration failed #3221 by @dcc123456 in #9821
• Feat: add SearXNG search tool to Agent (frontend + backend, i18n) by @RuyXu in #9699
• Fix: Optimized the style and functionality of multiple components #3221 #9817 by @dcc123456 in #9824
• Feat: Allow users to delete their profile pictures #3221 by @cike8899 in #9826
• Refa: revise the implementation of LightRAG and enable response caching by @yongtenglei in #9828
• Feat: Create a conversation before uploading files #3221 by @cike8899 in #9832
• UI updates by @writinwaters in #9836
• Fix: Optimize styling and add a search settings loading state #3221 by @dcc123456 in #9830
• Fix: Optimize page layout and style #3221 by @dcc123456 in #9852
• Feat: Move the dataset permission drop-down box to a separate file for better permission control #3221 by @cike8899 in #9850
• Docs: Refactored Retrieval component reference by @writinwaters in #9862
• Refactor: Improve the buffer close for vision_llm_chunk by @Woody-Hu in #9845
• Use 'float' explicitly for OpenAI's embedding "encoding_format" by @haiyuewa in #9838
• Fix: agent second round issue. by @KevinHuSh in #9863
• Fix: Fixed the issue where the agent and chat cards on the home page could not be deleted #3221 by @cike8899 in #9864
• Feat: By default, 50 records are displayed per page. #3221 by @cike8899 in #9867
• Feat: Initialize the data pipeline canvas. #9869 by @cike8899 in #9870
• Fix JSON serialization error for ModelMetaclass objects by @SaurabhLingam in #9812
• Fix: Optimize list display and rename functionality #3221 by @dcc123456 in #9875
• Add jemalloc install instruction for mac by @aopstudio in #9879
• Fix: The operator added by clicking the plus sign will overlap with the original operator. #9886 by @cike8899 in #9887
• Feat: add exponential back-off for Chat LiteLLM by @yongtenglei in #9880
• Docs: Updated the Code component reference by @writinwaters in #9884
• Feat: add canvas_category field for UserCanvas and CanvasTemplate by @yongtenglei in #9885
• Docs: Updated instructions on importing third-party packages to Sandbox by @writinwaters in #9890
• Update README by @yingfeng in #9904
• Feat: Allow users to enter SQL in the SQL operator #9897 by @cike8899 in #9898
• During the chat, the assistant's response cited documents outside current chat's kbs by @GrubbyLee in #9900
• Update API reference to use 'title' instead of 'name' for listing agents by @jayant-yadav in #9907
• Feat: Use storybook to display public components. #9914 by @cike8899 in #9915
• Fix: wrong chunk number while re-parsing document and keeping original chunks by @yongtenglei in #9912
• Feat: Display AvatarUpload and RAGFlowAvatar in Storybook #9914 by @cike8899 in #9920
• before docker-compose up, first down it,and cleans. by @zhanluxianshen in #9908
• Fix kb isolation infinity conn by @SyedShahmeerAli12 in #9913
• Feat: Added RenameDialog NumberInput and Spin storybook #9914 by @cike8899 in #9925
• fix: Optimize internationalization configuration #3221 by @dcc123456 in #9924
• Fix python_version in show_env.sh when its meets python3. by @zhanluxianshen in #9894
• fix xss in excel_parser by @zhanluxianshen in #9909
• Fix S3 client initialization with signature_version and addressing_style by @SyedShahmeerAli12 in #9911
• code cleans. by @zhanluxianshen in #9916
• Update .env by @kevinhonor in #9923
• Feat: Add metadata filtering function for /api/v1/retrieval by @huang-aoqin in #9877
• Refactor: Use re compile for weight method by @Woody-Hu in #9929
• Fix bugs for agent/tools. by @zhanluxianshen in #9930
• Feat: Allow users to select prompt word templates in agent operators. #9935 by @cike8899 in #9936
• Feat: Use sonner to replace the requested prompt message component #3221 by @cike8899 in #9951
• Feat: refine dataflow and initialize dataflow app by @yongtenglei in #9952
• Fix miscalculated token count by @biyuhao in #9776
• Refa: import issue. by @KevinHuSh in #9958
• add model: qwen3-max-preview by @TeslaZY in #9959
• Feat: Delete unused code in the data pipeline #9869 by @cike8899 in #9971
• Fix: The files in the knowledge base folder on the file management ...

v0.20.4

Summary

Improvements

• Agent component: Completes Chinese localization for the Agent component.
• Introduces the ENABLE_TIMEOUT_ASSERTION environment variable to enable or disable timeout assertions for file parsing tasks.
• Dataset:
  • Improves Markdown file parsing, with AST support to avoid unintended chunking.
  • Enhances HTML parsing, supporting bs4-based HTML tag traversal.

Added models

ZHIPU GLM-4.5

New Agent templates

Ecommerce Customer Service Workflow: A template designed to handle enquiries about product features and multi-product comparisons using the internal knowledge base, as well as to manage installation appointment bookings.

Fixed issues

• Dataset:
  • Unable to share resources with the team.
  • Inappropriate restrictions on the number and size of uploaded files.
• Chat:
  • Unable to preview referenced files in responses.
  • Unable to send out messages after file uploads.
• An OAuth2 authentication failure.
• A logical error in multi-conditioned metadata searches within a dataset.
• Citations infinitely increased in multi-turn conversations.

What's Changed

• Feat: Make the old page accessible via URL #3221 by @cike8899 in #9589
• Feat: reference should also be list after 0.20.x by @huang-aoqin in #9582
• Revert "Feat: reference should also be list after 0.20.x" by @KevinHuSh in #9592
• Fix (web): Update the style of segmented controls and add metallic texture gradients by @dcc123456 in #9591
• Fix:KeyError: 'globals' KeyError： 'globals' by @Woody-Hu in #9571
• fix:passing empty database array when updating assistant by @jiasu-hezhip in #9570
• Fix: Fixed the issue where the save button at the bottom of the chat page could not be displayed on small screens #3221 by @cike8899 in #9596
• Fix (web): Optimize text display effect #3221 by @dcc123456 in #9594
• Feat: Updated some colors according to the design draft #3221 by @cike8899 in #9599
• Fix: enlarge raptor timeout limits. by @KevinHuSh in #9600
• Feat: expand the capabilities of the MCP Server by @yongtenglei in #8707
• Feature (web): Optimize dataset pages and segmented components #3221 by @dcc123456 in #9605
• Feat: advanced markdown parsing by @yongtenglei in #9607
• Fix: Remove the file size and quantity restrictions of the upload control #9613 #9598 by @cike8899 in #9618
• Fix: duplicated role... by @KevinHuSh in #9622
• Fix (style): Optimized Datasets color scheme and layout #3221 by @dcc123456 in #9620
• Feat: Allow users to parse documents directly after uploading files #3221 by @cike8899 in #9633
• Chore: Update infinity-sdk from 0.6.0.dev4 to 0.6.0.dev5 by @asiroliu in #9628
• Fixes (web): Optimized search page style and functionality #3221 by @dcc123456 in #9627
• fix(dataset, next-chats): Fix data form data acquisition logic And Optimize the chat settings interface and add language selection by @muzilib in #9629
• Fix: The buttons at the bottom of the dataset settings page are not visible on small screens #9638 by @cike8899 in #9639
• Fix: Fixed an issue where knowledge base could not be shared #9634 by @cike8899 in #9642
• Update fr.ts - RAPTOR Issue prompt by @Konsilion in #9646
• fix(sdk): add default empty dict for metadata_condition by @likenamehaojie in #9640
• Refactor: based on async await to handle Redis when raptor by @Woody-Hu in #9576
• Fix: Document Previewer is not working #9606 by @cike8899 in #9656
• Refactoring: Agent completions API change response structure by @huang-aoqin in #9631
• Fix: embedded search AI summary by @yongtenglei in #9658
• Fix: Display the invited icon in the header #9634 by @cike8899 in #9659
• Fix: code executor timeout by @yongtenglei in #9671
• Fix: Search app AI summary ERROR And The tag set cannot be selected #9649 #9652 by @dcc123456 in #9664
• Fix: unexpected LLM parameters by @yongtenglei in #9661
• Fix: Place the invitation reminder icon in a separate file #9634 by @cike8899 in #9662
• Fix: meta data error. by @KevinHuSh in #9670
• Fix:[ERROR]'LLMBundle' object has no attribute 'language' by @Woody-Hu in #9682
• Fix: Wrap VersionDialog in DropdownProvider for proper context by @uciadonis in #9677
• Feat: Exclude operator_permission field from renaming chat fields #3221 by @cike8899 in #9692
• Fix: grow reference list by @huang-aoqin in #9674
• fix: Create conversation sessions will lost prologue by @huang-aoqin in #9666
• Refa: improve lightrag by @yongtenglei in #9690
• Fix: Optimize dataset page layout and internationalization and default values for multi selection #3221 by @dcc123456 in #9695
• Refa: replace Chat Ollama implementation with LiteLLM by @yongtenglei in #9693
• fix: validate chunk type before processing to prevent AttributeError by @Mckennasora in #9698
• Docs: update API endpoint paths by @asiroliu in #9683
• Fix: meta data filter with AND logic operations. by @KevinHuSh in #9687
• Fix: Delete the uploaded file in the chat input box, the corresponding file ID is not deleted #9701 by @cike8899 in #9702
• Fix: Optimize table style #3221 by @dcc123456 in #9703
• Refactor: for total_token_count method use if to check first. by @Woody-Hu in #9707
• Fix: Chunk error when re-parsing created file #9665 by @dcc123456 in #9711
• Update llm_factories.json by @ycz87 in #9714
• Feat: add Zhipu GLM-4.5 model series by @yongtenglei in #9715
• Fix: Try to fix the issue of not being able to log in through Oauth2 #9601 by @cike8899 in #9717
• Fix: After deleting the knowledge graph, jump to the dataset page #9722 by @cike8899 in #9723
• Fix(i18n): Added new translations #3221 by @dcc123456 in #9727
• Docs: Miscellaneous updates by @writinwaters in #9729
• Fix: add mode for embeded agent. by @KevinHuSh in #9741
• Fix: Optimize tooltips and I118n #3221 by @dcc123456 in #9744
• Feat: Upgrade html parser by @pingguoCooler in #9675
• Fix: Fixed the issue that the agent embedded page needs to be logged in #9750 by @cike8899 in #9751
• Fix: Optimize the MultiSelect component and system prompt templates #3221 by @dcc123456 in #9752
• Create ecommerce_customer_service_workflow.json by @YyBoom233 in #9755
• Added v0.20.4 release notes by @writinwaters in #9757
• Docs: Update version references to v0.20.4 in READMEs and docs by @asiroliu in #9758
• Revert: broken agent completion by #9631 by @yongtenglei in #9760
• Fix: search list permission by @yongtenglei in #9767

New Contributors

• @Konsilion made their first contribution in #9646
• @likenamehaojie made their first contribution in #9640
• @uciadonis made their first contribution in #9677
• @ycz87 made their first contribution in #9714
• @pingguoCooler made their first contribution in #9675

Full Changelog: v0.20.3...v0.20.4