✨ Enhancements

• Add count field and support non-ASCII characters in DataScientist action output (#2812)
• support knowledgeSpace custom retrieve strategy (#2842)
• AI/ML API integration (#2844)
• update client usage to prefer await aclose() in async context" (#2839)
• Add GaussDB data source connector (#2813)
• add storage client cache (#2810)
• Support GLM4.1-vl model (#2806)
• Support export messages (#2802)
• Add operator for structured output data of datasource in AWEL (#2794)
• Support MLX inference (#2781)
• Support Qwen3 embeddings (#2772)

🐞 Bug fixes

• resolve KeyError and validation issues with custom prompt templates (#2848)
• improve JSON parsing robustness in output parser (#2850)
• Fix newline escape issue (#2854)
• Update Milvus truncate method for GraphRAG (#2847)
• resolve overlapping floating buttons in web (#2836)
• update client usage to prefer await aclose() in async context" (#2839)
• Fix auto-scroll issue in chat (#2797)
• Fix milvus vector_name_exists bug (#2818)
• Fix SummaryExtractor not found error (#2805)
• fix v2 api error message (#2801)
• prompt selection and prompt_code delivery for chat_dashboard (#2790)
• optimize chat message auto scroll (#2792)
• fix checkBox for singleAgent mode (#2785)
• Remove SQL comments before formatting the SQL (#2769)
• Fix proxy reranker bug (#2783)
• Change plan manager prompt in order to avoid calling tools directly (#2778)

🛠️ Other improvements

• doc: fix ob rag doc (#2840)
• update CONTRIBUTING.md to use uv package manager (#2855)
• update create chroma store param (#2798)
• update 0.7.3 version (#2860)

👥 New Contributors

• @D1m7asis
• @HYP-hu
• @huhao0926
• @jiao-duan
• @myworking2012
• @stevenlx96

Thank you to all our contributors for making this release possible!
@Aries-ckt, @D1m7asis, @HYP-hu, @WangzJi, @chenliang15405, @fangyinc, @geebytes, @huhao0926, @jiao-duan, @myworking2012, @stevenlx96 and @vnicers

中文发版说明： https://www.yuque.com/eosphoros/dbgpt-docs/wfwcwqldi9hkusvb

✨ Enhancements

• Support MiniCPM4-8B (#2754)
• Support Qwen2.5 VL models (#2743)
• Enhance dashboard visualization and table-chart formatting (#2731)
• Add hybrid search for knowledge space. (#2722)
• Add chat_dashboard support to V2 API (#2726)
• Dashboard supports Pie Chart visualization (#2709)
• Add Chat Data template (#2716)
• Add chat message auto scroll (#2682)
• Support llama.cpp for GLM4-z1 (#2690)
• Add patch for Qwen3 moe (#2676)
• Add tree retriever based on document directory level (#2669)

🐞 Bug fixes

• style: fix readme style (#2762)
• Fix agent retriever resource (#2719)
• Bugfix(RAG):handle exceptions in aload_document_with_limit results (#2712)
• Resolve warnings of logger library (#2681)
• correct argument limit error (#2758)
• Initialize prompt templates to ensure all templates are load (#2729)
• fix unit test and some code typos (#2717)
• fix chat scene config custom prompt (#2725)
• Fix typo (#2696)
• resolve KeyError: 'question' in chat_dashboard prompt template (#2695)
• issue 2685 (#2686)
• fix task concurrency and batch processing issue (#2671)
• fix(VectorStore) fix MilvusStore to use serialize function for metadata json encoding (#2672)

🛠️ Other improvements

• update readme style (#2767)
• fix div style (#2761)
• update readme (#2760)
• Update dependency reference path (#2745)
• Update ding group (#2724)
• Remove python-dotenv (#2721)
• upgrade pandas version (#2705)
• integration oracle doc (#2701)

Thank you to all our contributors for making this release possible!
@Aries-ckt, @FOkvj, @WangzJi, @chenliang15405, @china-zhz, @cumbersun, @emmanuel-ferdman, @fangyinc, @geebytes, @vnicers and tam

✨ Enhancements

• Optimize agent memory (#2665)
• The SQL execution result of the DataScientist can be passed to the next agent (#2654)
• Support Qwen3 models (#2664)
• add model provider InfiniAI (#2653)
• add oracle datasource (#2629)
• Proxy multimodal supports (#2641)
• Support GLM-4-0414 (#2630)
• Multimodal supports (#2627)
• feat(mode) : Adding the siliconflow embeding proxy (#2603)
• Support SSL/TLS for MCP (#2591)
• Add git for docker image (#2584)
• Support MCP authentication (#2582)
• Compatible with MCP and tools (#2566)
• More general ReAct Agent (#2556)
• Add Word97-2003 (.doc) Binary File parsing module (#2544)
• Support html/svg preview (#2518)
• add Text Embeddings Inference (TEI) API support for reranking (#2516)

🐞 Bug fixes

• fix remote reranker api & fix long_term memory (#2648)
• Fix long-term memory error (#2644)
• fix(reranker) The rerank model is used during the knowledge base recall for chat scenarios and recall test scenarios (#2638)
• fix connect mssql embedding error (#2589)
• storage add collection bug (#2615)
• update doc example code to work (#2624)
• Fix load too many chunks error (#2614)
• the issue is the app_code can't be stored in the chat_history… (#2606)
• Fix apiserver error (#2605)
• change flow_data column type from text to longtext (#2590)
• fix ReActAgent (#2588)
• Fix resource binding bug (#2572)
• Fix KB document questions list check logic (#2565)
• Prevent dragging node when selecting text in parameter input fields (#2560)
• make tugraph work again (#2551)
• fix vllm tokenizer count error (#2555)
• some typos using typos (#2552)
• issue 2549 (#2550)
• clouse issue #2469 by fixing the type (#2533)
• wrong name close #2530 (#2532)

🛠️ Other improvements

• Fix build doc error (#2517)
• Add 0.7.0 release blog (#2515)
• [BUG] SQL Injection through CVE Bypass in DB-GPT 0.7.0 (CVE-2024-10835 & CVE-2024-10901) (#2650)
• [BUG] RCE Vulnerability in DB-GPT Plugin Upload System (#2649)
• Fix Clickhouse SQL syntax error for RdbmsSummary (#2651)
• Fix postgresql (#2601)
• Bugfix/elasticsearch store read config dict (#2577)
• Fix ChatWithDbQA param validate (#2569)
• doc: fix a typo in README (#2526)

👥 New Contributors

• @cumbersun made their first contribution in #2550
• @chenliang15405 made their first contribution in #2569
• @lijun51888 made their first contribution in #2606
• @dobet made their first contribution in #2589
• @seoibiubiu made their first contribution in #2651
• @luowenwu made their first contribution in #2629
• @paxionfull made their first contribution in #2653
• @geckosecurity made their first contribution in #2649

Thank you to all our contributors for making this release possible!
@Aries-ckt, @FOkvj, @chenliang15405, @csunny, @cumbersun, @dobet, @fangyinc, @geckosecurity, @geebytes, @lijun51888, @luowenwu, @paxionfull, @seoibiubiu, @vnicers, @yihong0618, @yyhhyyyyyy and tam

中文发版说明： https://www.yuque.com/eosphoros/dbgpt-docs/orsg16hwke26i6f8

✨ Enhancements

• Supports MCP (#2497)
• Support model icon & i18n (#2486)
• OSS&S3 Supports (#2477)
• Support simple templates (#2463)
• Agent fixes and parameter additions (#2457)
• Support docker install (#2442)
• Better Chat Excel (#2423)
• add ReActAgent (#2420)
• API support reasoning (#2409)
• Support reasoning for ChatDashboard (#2401)
• Support reasoning model (#2375)
• Database connection Renewal (#2359)
• feature(agent): Knowledge add retriever name&desc getter and stter (#2309)
• add app starter role in mutli agent (#2265)
• Support llama.cpp server deploy (#2263)
• Support Gitee models (#2257)
• fix agent overrid bug (#2235)

🐞 Bug fixes

• Modify AWEL init log param (#2510)
• rag knowledge process workflow template (#2509)
• milvus collection search param (#2481)
• fix make test failed (#2502)
• Fix param cache error (#2500)
• column alias bug fix (#2499)
• doris_conn cannot get table comment to store vectordb (#2490)
• issue2484 (#2488)
• issue2484 (#2485)
• handle empty response in LLM extractor (#2480)
• rag storage refactor (#2434)
• Add system_app parameter to knowledgeSpaceRetriever initialization (#2472)
• Explicitly create data tables from the df (#2437) (#2464)
• fix (vector store) Database refresh error, vector collection deduplication consistency issue (#2465)
• Fix AWEL flow error (#2461)
• Correct the import path for StarRocks dialect (#2456)
• Fix build arm image bug (#2446)
• aembed_query type error (#2441)
• Fix chat completions API error (#2443)
• fix issue2321:Provide an implementation of the truncate method (#2357)
• move app resource implement to serv pkg (#2310)
• Fix build lyric error (#2413)
• Fix reasoning output bug (#2405)
• solve default persist path (#2395)
• Fix datasource resource error (#2394)
• Fix reasoning output bug (#2393)
• resolve clickhouse connection error (#2380)
• Fix local embedding error (#2371)
• Fix read i18n config error (#2368)
• agent's llmclient bug (#2298)
• issue2323 (#2325)
• fix function find_json_objects (#2289)
• Fix load db models error (#2290)
• remove invaild file (#2278)
• Fix build docker image error (#2279)
• prevent SQL injection in chart data query (CVE-2024-10901) (#2269)
• prevent arbitrary file write in SQL editor (CVE-2024-10835) (#2268)
• format problem (#2275)
• Fix push document image error (#2273)
• Fix document build error (#2271)
• fix path traversal vulnerability (CVE-2024-10834) (#2267)

🛠️ Other improvements

• Code format (#2513)
• Add 0.7.0 workflow (#2493)
• add vscode devcontainer config (#2466)
• rag storage refactor (#2434)
• add ollama docs (#2419)
• fix unit test (#2421)
• add ollama config and support ollama model output (#2411)
• Refactor dbgpts for 0.7.0 (#2397)
• add vllm llama_cpp docs and standardize configs (#2386)
• Fix build doc error (#2385)
• Fix build doc error (#2383)
• Config documents & i18n supports (#2365)
• adapt rag storage and add integration documents. (#2361)
• restructure modules and config handling (#2358)
• add disckaimer (#2274)
• wechat update (#2270)
• deps(tongyi): fix tongyi dependencies and add tongyi proxy config (#2467)
• workflow(ruff): add ruff code checks workflow (#2438)
• GraphRAG: refine config usage and fix some bug. (#2392)
• doc:add integrations documents (#2382)
• deps:refactor dependencies installation method (#2381)
• GraphRAG: add new feature and bugfix (#2373)
• add LOCAL_DB_SSL_VERIFY for Supporting TiDB serverless (#2308)
• Support text2gql search for GraphRAG (#2227)
• Support embedding similarity search for GraphRAG (#2200)
• Fixed agent&awel examples (#2256)
• Fix Bug #2224 (#2240)

Thank you to all our contributors for making this release possible!
283569391@qq.com, @15089677014, @Aries-ckt, @FOkvj, @Jant1L, @SonglinLyu, @TenYearOldJAVA, @Weaxs, @cinjoseph, @csunny, @damonqin, @dusx1981, @fangyinc, @geebytes, @haawha, @utopia2077, @vnicers, @xuxl2024, @yhjun1026, @yunfeng1993, @yyhhyyyyyy and tam

Chinese Release Notes: https://www.yuque.com/eosphoros/dbgpt-docs/asweou4i9rhnwchm

✨ Enhancements

• Add release pipy workflow (#2222)
• Agent supports conversation context (#2230)
• Support max output tokens parameter (#2218)
• Add Knowledge Process Workflow (#2210)
• Support siliconflow rerank models (#2188)
• Vector Data visualize for Chat Data (#2172)
• Support concurrent community summarization (#2160)
• Support siliconflow models (#2157)
• Support claude proxy models (#2155)
• Feat rdb summary wide table (#2035)

🐞 Bug fixes

• Upgrade lyric version (#2229)
• Fix awel template css error (#2220)
• Fix create same graph name error. (#2215)
• Fix build docker image error (#2221)
• Limit the version of fastapi (#2211)
• Fix the necessary resource for the action in error log (#2197)
• Fix httpx v0.28.0 proxies bug (#2179)
• Fix bug fail to delete files when deleting the knowledge space (#2185)
• Fix sql not found error for chat data (#2152)
• Fix VectorStore can not be attached to EmbeddingAssemblerOperator bug (#2170)

🛠️ Other improvements

• Add DB-GPT Agent paper (#2232)
• Add oss url for tugraph image (#2228)
• Upgrade dbgpt-tugraph-plugins to version 0.1.1 (#2213)
• Update TuGraph database version to 4.5.0 (#2209)
• Modify siliconflow name (#2194)

Thank you to all our contributors for making this release possible!
@Appointat, @Aries-ckt, @FOkvj, @GITHUBear, @HYSMN, @Sween1y, @fangyinc, @fanzhidongyzby, @toralee and @yhjun1026

✨ Enhancements

• Distinguish awel flow on the ui (#2149)
• Pdf loader support table extraction (#2147)
• Support Qwen2.5 coder models (#2139)
• Feat/note book (#2134)
• Call xunfei spark with stream, and fix the temperature bug (#2121)
• Execute codes in a sandbox environment (#2119)
• Feat support xunfei spark ultra/max/pro/lite (#2117)
• Enhance the triplets extraction in the knowledge graph by the batch size (#2091)

🐞 Bug fixes

• Fix the issue that lost prev observation memory in toolExpert bug (#2136) (#2137)
• Fix core agent dependent prompt service bug (#2129)
• Fix milvus store search with topk always set to 4 (#2131)
• Fix use fuzzy matching when searching dbgpts (#2110)
• Fix the issue that OceanBase config in .env does not take effect (#2108)
• os environment variables should take precedences over the defaults, f… (#2100)
• Fix Critical Directory Traversal Vulnerability (#2098)
• Fix excel_reader.py unicode bug (#2095)

🛠️ Other improvements

• v0.6.2 tag (#2148)
• Add AWEL shared slides (#2107)

Thank you to all our contributors for making this release possible!
@Appointat, @Aries-ckt, @GITHUBear, @ayushin, @cinjoseph, @fangyinc, @fengdezhen666, @hiyizi, @jiaoqiyuan, @r0path, @slper001, @sunshinesmilelk and @yhjun1026

✨ Enhancements

• Passing stop parameter to proxyllm (#2077)
• Add document structure into GraphRAG (#2033)
• Update OceanBaseStore to be compatible with oceanbase-4.3.3.0 (#2074)
• Add evaluation service module for RAG and Agent (#2070)
• Support Qwen2.5 models (#2030)

🐞 Bug fixes

• Fix unit test error (#2085)
• Fix chunk strategy (#2075)
• Replace deprecated method and add missing decorator (#2072)
• Fixed the error caused by file format when docx loading files (#2061)
• Resolving duplicate dependency (#2055)
• Fix long message response error (#2051)
• cli: issue close strategy update (#2037)
• Fix vllm inference error (#2025)
• fix bug - Tongyiproxy embedding model error (#2018)

🛠️ Other improvements

• refinement of document structure process (#2084)
• app & agents development document (#2081)
• adjust graphrag doc (#2027)

Thank you to all our contributors for making this release possible!
@517739, @Appointat, @Aries-ckt, @GITHUBear, @Hopshine, @KingSkyLi, @csunny, @fangyinc, @fanzhidongyzby, @iCanDoAllThingszz, @lhwan, @mzaispace, @nightosong and @noahc1510

✨ Enhancements

• Support higher-order operators (#1984)
• Add DAG variables to web flow (#1981)
• Add GraphRAG web statics (#1943)
• ✨ feat(GraphRAG): enhance GraphRAG by graph community summary (#1801)
• Unified frontend code style (#1923)
• Support qianfan embedding and resolves an issue (#1819)
• New dbgpts modules (#1910)
• AWEL flow 2.0 frontend codes (#1898)
• AWEL flow 2.0 backend code (#1879)
• Web update (#1860)
• (0.6)New UI (#1855)
• Add financial report how to use doc. (#1757)
• Support mistral nemo (#1766)
• Add Text2GQL fine tuning framework and provide TuGraph examples (DB-GPT-Hub#287)

🐞 Bug fixes

• Fix build doc image error (#1997)
• Fix export flow error (#1994)
• Fix full text storage truncate method error. (#1989)
• Fix native app bug (#1988)
• Fix agent bug (#1953)
• rename vis-chart to vis-db-chart (#1946)
• Fix postgresql read schema_info with characters truncated if too long (#1935)
• Fix chat web bug (#1932)
• Fix dbgpts modules bug (#1921)
• Fix execution failure error after manually updating SQL (#1812)
• Fix Graph retrieve bug (#1884)
• Fix can't read the milvus url config (#1808)
• Fix pylint check error (#1807)
• Unique Constraint Issue with create_space API Cau… (#1794)
• add ding group & update wechat (#1787)
• Fix knowledge api report error when call create space api (#1784)
• Editor page with redundant table fields of the same name in othe… (#1765)
• Resolve async response issue with Azure GPT-4o API (#1772)
• Fix knowledge command load bug (#1767)
• Fix db schema aretriever bug (#1755)

🛠️ Other improvements

• Add v0.6.0 tag (#2000)
• Add flow usage document (#1999)
• Add v0.6.0 release notes (#1991)
• Build flow UI (#1992)
• 🍎 docs: add docs for v0.6.0 (#1964)
• Update Graph RAG Document (#1955)
• Build static file (#1951)
• Fix pylint error (#1915)
• Upgrade gpt-vis (#1883)

Thank you to all our contributors for making this release possible!
@Aries-ckt, @Dreammy23, @Hec-gitHub, @JxQg, @KingSkyLi, @M1n9X, @bigcash, @chaplinthink, @csunny, @dusens, @fangyinc, @fanzhidongyzby, @huangjh131, @hustcc, @lhwan, @whyuds, @yhjun1026 and @yunfeng1993

✨ Enhancements

• Support Financial Report Analysis (#1702)
• Support llama3.1 models (#1744)
• Multiple ways to run dbgpts (#1734)
• Support internlm2.5 models (#1735)
• Support codegeex4-all-9b (#1720)
• Support opentelemetry exporter (#1690)

🐞 Bug fixes

• Fix chat knowledge error (#1753)
• Fix the problem of invalid subgraph naming when creating tugraph (#1731)
• Fix line chart using interval field as xAxis & fix pie chart tooltip (#1697)
• Add missing context property (#1732)
• Fix table scroll bug (#1725)
• Fix Stream object can't use await (#1719)
• Fix http empty headers error (#1710)
• FIx GraphRAG build error (#1706)
• Fix write log files bug (#1687)

🛠️ Other improvements

• Multi-version document build (#1741)

Thank you to all our contributors for making this release possible!
@A-Salty-Fish, @Aries-ckt, @Kain-90, @chenluli, @fangyinc and @shyusi

✨ Enhancements

• Support gemma-2 model (#1675)
• Support tongyi embedding (#1552)
• Support database model registry (#1656)
• Add download image function&fix null data point (#1653)
• Support sort line chart data according to time order (#1639)

🐞 Bug fixes

• Fix bug for issue #1669 (#1672)
• Fix web build issues with yarn and missing dependencies (#1658)
• Fix AWEL branch bug (#1640)
• Fix schema linking error (#1637)

🛠️ Other improvements

• tag_v0.5.9 (#1674)
• Add agents development guide (#1673)

Thank you to all our contributors for making this release possible!
@Aries-ckt, @Armandwt, @aagnone3, @chenluli, @fangyinc and @zhenchaozhu