v0.4.0.dev13

What's new

• An initial version of the migration guide is ready. Find it here! (#4765)
• Model family is now available in the model client (#4856)

Breaking changes

• Previously deprecated module paths have been removed (#4853)
• SingleThreadedAgentRuntime.process_next is now blocking and has moved to be an internal API (#4855)

Fixes

• Fix SingleThreadedAgentRuntime busy loop by @jackgerrits in #4855
• Fix BaseOpenAIChatCompletionClient token usage by @gziz in #4770

Doc changes

• Migration guide for 0.4 by @ekzhu in #4765
• Clarify tool use in agent tutorial by @ekzhu in #4860
• AgentChat tutorial update to include model context usage and langchain tool by @ekzhu in #4843
• Add missing model context attribute by @Leon0402 in #4848

Other

• Replace Tuple[type[ChatMessage], ...] with Sequence[type[ChatMessage]] by @ekzhu in #4857
• Make register_factory a user facing API by @jackgerrits in #4854
• Move intervention objects to root module by @jackgerrits in #4859

New Contributors

• @Leon0402 made their first contribution in #4848

Full Changelog: v0.4.0.dev12...v0.4.0.dev13

v0.4.0.dev12

Important Changes

• run and run_stream now support a list of messages as task input.
• Introduces AgentEvent union type in AgentChat, for all messages that are not meant to be consumed by other agents. Replace AgentMessage with AgentEvent | ChatMessage union type in your code, e.g., in your custom selector function for SelectorGroupChat and processing code for TaskResult.messages.
• Introduce ToolCallSummaryMessage to ChatMessage for tool call results from agents. Read AssistantAgent Doc
• Introduce ModelContext parameter for AssistantAgent, allow usage of BufferedChatCompletionContext to limit context window size sent to model.
• Introduce ComponentConfig and add configuration loader for ChatCompletionClient. See Component Config
• Moved autogen_core.tools.PythonCodeExecutorTool to autogen_ext.tools.code_execution.PythonCodeExecutionTool.
• Documentation updates.

Upcoming Changes

• Deprecating @message_handler. Use @event or @rpc to annotate handlers instead. @message_handler will be kept with a deprecation warning until further notice. #4828
• Token counting mechanism bug fixes #4719

New Contributors

• @hsm207 made their first contribution in #4655
• @iamarunbrahma made their first contribution in #4500
• @inbal2l made their first contribution in #4717
• @r-bit-rry made their first contribution in #4759
• @jspv made their first contribution in #4755
• @akurniawan made their first contribution in #4681
• @lanbaoshen made their first contribution in #4767
• @srjoglekar246 made their first contribution in #4801
• @richard-gyiko made their first contribution in #4826
• @kimmywork made their first contribution in #4732

Full Changelog: v0.4.0.dev11...v0.4.0.dev12

v0.2.40

What's Changed

• Add warning message when NoEligibleSpeaker by @thinkall in #4535
• [Bug]: Bedrock client uses incorrect environment variables for authentication by @vaisakh-prod in #4657
• fix "keep_first_message" to make sure messages are in correct order, by @milkmeat in #4653
• fix: No context vars for async agents replies by @violonistahiles in #4640
• update version by @ekzhu in #4713

New Contributors

• @vaisakh-prod made their first contribution in #4657
• @milkmeat made their first contribution in #4653
• @violonistahiles made their first contribution in #4640

Full Changelog: v0.2.39...v0.2.40

v0.4.0.dev11

Important Changes:

1. autogen_agentchat.agents.AssistantAgent behavior has been updated in dev10. It will not perform multiple rounds of tool calls -- only 1 round of tool call, followed by an optional reflection step to provide a natural language response. Please see the API doc for AssistantAgent.
2. Module renamings:

• autogen_core.base --> autogen_core
• autogen_core.application.SingleThreadedAgentRuntime --> autogen_core.SingleThreadedAgentRuntime.
• autogen_core.application.logging --> autogen_core.logging.
• autogen_core.components.* --> autogen_core.*, for models, model_context, tools, tool_agent and code_executor.
• autogen_core.components.code_executor.LocalCommandLineCodeExecutor --> autogen_ext.code_executors.local.LocalCommandLineCodeExecutor
• autogen_agentchat.task.Console --> autogen_agentchat.ui.Console.
• autogen_agentchat.task.<termination conditions> --> autogen_agentchat.conditions.<termination_conditions>.
• autogen_ext.models gets divided into autogen_ext.models.openai, autogen_ext.models.replay.
• autogen_ext.agents gets divided into separate submodules for each extension agent class.
• autogen_ext.code_executors gets divided into separate submodules for each executor class.

New Contributors

• @eranco74 made their first contribution in #4639

Full Changelog: v0.4.0.dev9...v0.4.0.dev11

v0.2.39

What's Changed

• fix: GroupChatManager async run throws an exception if no eligible speaker by @leryor in #4283
• Bugfix: Web surfer creating incomplete copy of messages by @Hedrekao in #4050

New Contributors

• @leryor made their first contribution in #4283
• @Hedrekao made their first contribution in #4050

Full Changelog: v0.2.38...v0.2.39

v0.2.38

What's Changed

• docs: fix bullet formatting for kubernetes pod executor by @lukehsiao in #3911
• Fix 0.2 Quickstart example llm_config issue by @victordibia in #4060
• Bugfix: Optional rate limiting by @kampernet in #4066

New Contributors

• @lukehsiao made their first contribution in #3911
• @kampernet made their first contribution in #4066

Full Changelog: v0.2.37...v0.2.38

v0.2.37

What's Changed

• Use trusted publisher for pypi release by @jackgerrits in #3596
• Fix typos in Cerebras doc by @henrytwo in #3590
• Add blog post announcing the new architecture preview by @jackgerrits in #3599
• Update PR link in blog post by @jackgerrits in #3602
• Create CI to tag issues with needs triage by @jackgerrits in #3605
• Update issue templates by @jackgerrits in #3610
• Fix small typo in the docs by @jknaudt21 in #3650
• Update 0.2 CI to target branch, remove merge queue by @jackgerrits in #3656
• Update BaseUrl of docusaurus site by @jackgerrits in #3658
• Add announcement bar for 0.4 by @jackgerrits in #3717
• Update links on 0.2 website by @jackgerrits in #3734
• Function Calling Support for Gemini - Part 2 by @luxzoli in #3726
• Fix #2643 - groupchat model registration by @Matteo-Frattaroli in #2696
• Added a demonstartion notebook featuring the usage of Langchain with AutoGen by @Kirushikesh in #3461
• Autobuild Function calling by @krishnashed in #3238
• Update Docs to Point to 0.4 by @victordibia in #3764
• Notebook on web crawling by @WilliamEspegren in #2720
• Fix link to v0.4 documentation by @ekzhu in #3772
• Remove path filter for website testing in 0.2 by @jackgerrits in #3782
• Fix broken image URL in README by @gagb in #3776
• Clarify stable package name and version on home page by @jackgerrits in #3775
• Fix CTA button alignment in docs home page by @victordibia in #3788
• K8s code executor by @questcollector in #3419
• Add Couchbase Vector DB Example Notebook and Minor Bug Fix by @lokesh-couchbase in #3804
• Add Zep ecosystem doc and notebook by @danielchalef in #3681
• [bug] Changes Text Cache Default to None by @WaelKarkoub in #3872
• [bug] Validates If The Role Tool is Handled Correctly after Transforms by @WaelKarkoub in #3875
• [CAP] Abstraction of actor_connector to go along with runtime factory and runtime abstraction by @rajan-chari in #3296

New Contributors

• @jknaudt21 made their first contribution in #3650
• @Matteo-Frattaroli made their first contribution in #2696
• @WilliamEspegren made their first contribution in #2720
• @questcollector made their first contribution in #3419
• @danielchalef made their first contribution in #3681

Full Changelog: v0.2.36...v0.2.37

v0.2.36

Important

In order to better align with a new multi-packaging structure we have coming very soon, AutoGen is now available on PyPi as autogen-agentchat as of version 0.2.36.

pip install autogen-agentchat~=0.2

Highlights

• Ability to add MessageTransforms to the GroupChat's Select Speaker nested chat (speaker_selection_method='auto') by @marklysze in #2719
• Integrate Mem0 for providing long-term memory for AI Agents by @Dev-Khant in #3389
• Amazon Bedrock Client for AutoGen by @Hk669 in #3232
• Add Couchbase VectorDB Support by @lokesh-couchbase in #3525
• Portkey Integration with Autogen by @siddharthsambharia-portkey in #3395
• Add Cerebras Integration by @henrytwo in #3585
• Ollama Client (with tool calling) by @marklysze in #3056
• Ollama library v0.3.3 update with changes to options handling by @marklysze in #3593

What's Changed

• [.Net] Dispose kernel after running dotnet interactive tests by @LittleLittleCloud in #3378
• Ensure 'name' on initial message by @marklysze in #2635
• [.Net] Enable step-by-step execution for two-agent conversation SendAsync API by @LittleLittleCloud in #3360
• Add contributor list via contributors.md by @ekzhu in #3384
• Update CONTRIBUTORS.md; fix cell order by @gagb in #3386
• Update CONTRIBUTORS.md; fix broken URL by @gagb in #3387
• Update AutoGen.Net Contributors.md by @LittleLittleCloud in #3393
• Update CONTRIBUTORS.md by @rickyloynd-microsoft in #3391
• Add Language Agent Tree Search (LATS) notebook by @andyz245 in #3376
• [.Net] Release 0.1.0 by @LittleLittleCloud in #3398
• [.Net] Rename AutoGen.OpenAI to AutoGen.OpenAI.V1 by @LittleLittleCloud in #3358
• Update Docker.md；fix broken URL by @New-World-2019 in #3399
• Fix embedding function assignment in QdrantVectorDB by @eddyfidel in #3396
• Add mongodb to topic guide docs by @thinkall in #3400
• Ability to add MessageTransforms to the GroupChat's Select Speaker nested chat (speaker_selection_method='auto') by @marklysze in #2719
• Fix for group chat resume - full chat history for each agent by @marklysze in #3412
• Update agent_chat.md；Fix broken URL by @New-World-2019 in #3416
• Add None back to remove_termination_string for a_resume in group chat by @heyitsaamir in #3410
• Amazon Bedrock Client for AutoGen by @Hk669 in #3232
• fix ImportFromModule is unhashable issue by @zcipod in #3362
• Transform to add an agent's name into the message content by @marklysze in #3334
• [Gallery] Adding a new Finance Tracker Demo Application by @Kirushikesh in #3414
• Update contributors by @davorrunje in #3420
• Update privacy link in readme and add Consumer Health Privacy notice on website by @jackgerrits in #3422
• Add studio pre-print by @gagb in #3423
• Update README.md by @gagb in #3424
• Update CITATION.cff by @gagb in #3427
• Add details about GAIA benchmark evaluation by @jackgerrits in #3433
• Add missing contributors by @gagb in #3426
• Update package.json, remove gh-pages dep by @victordibia in #3435
• [.Net] Add AutoGen.OpenAI package that uses OpenAI v2 SDK by @LittleLittleCloud in #3402
• Autogenstudio Update - Support for Anthropic/Mistral, Other Updates by @victordibia in #3439
• Fix AutoGen Studio pathPrefix error on Windows by @New-World-2019 in #3406
• fix: tool calling cohere by @Anirudh31415926535 in #3355
• [.Net] add output schema to generateReplyOption by @LittleLittleCloud in #3450
• Use utf-8 encoding at subprocess.run() by @giorgossideris in #3454
• fix model_config SQLModel by @Hk669 in #3457
• Fix for Anthropic client class so system messages aren't lost by @marklysze in #3392
• docs: fix autogen install url by @Lulalaby in #3459
• no more then -> no more than #typo by @scenaristeur in #3469
• [Graph RAG] Init Commit with GraphRag interfaces by @randombet in #3388
• Update usage of built-in LLMs in Microsoft Fabric by @thinkall in #3474
• [.Net] In AutoGen.OpenAI and AutoGen.OpenAI.V1, stop setting name field when assistant message contains tool call by @LittleLittleCloud in #3481
• [.Net] release v0.2.0 by @LittleLittleCloud in #3483
• Integrate Mem0 for providing long-term memory for AI Agents by @Dev-Khant in #3389
• Update readme by @gagb in #3485
• Ecosystem Mem0: change title by @Dev-Khant in #3492
• Support for Human Input Mode in AutoGen Studio by @victordibia in #3484
• Fix the bug of creating a new session on the AutoGen Studio playgroun… by @New-World-2019 in #3407
• Portkey Integration with Autogen by @siddharthsambharia-portkey in #3395
• Update pyproject.toml for AutoGen Studio by @gagb in #3495
• Update README.md by @gagb in #3499
• [.Net][AutoGen.OpenAI] Allow nullable system message, temperature and max token to support o1-preview model by @LittleLittleCloud in #3524
• [.NET] Release v0.2.1 by @DavidLuong98 in #3529
• Fix issue caused by chromadb error type change by @thinkall in #3557
• Bump express from 4.19.2 to 4.21.0 in /website by @dependabot in #3552
• typo intiate_chat -> initiate_chat by @scenaristeur in #3533
• #3538: issue: 'https://docs.trychroma.com/embeddings' hyperlink is … by @CellCS in #3539
• Fix syntax error in user-defined-functions docs (unecessarily capital… by @SuchitG04 in #3336
• Bump webpack from 5.89.0 to 5.94.0 in /website by @dependabot in #3438
• Update LLM Caching link in Optional-Dependencies.md by @evanhadfield in #2885
• Fix for async group chat resume - full chat history for each agent by @hardchor in #3558
• fix: update cohere tool calling multi agents by @Anirudh31415926535 in #3488
• Fix role mapping in GPTAssistantAgent for OpenAI API compatibility by @evandavid1 in #3555
• Fix Exception in a_generate_reply when function_call is None by @artero in #3250
• Catch token count issue while streaming with customized models by @BeibinLi in #3241
• Fix AttributeError: 'NoneType' object has no attribute 'name' by @drewgillson in #3127
• Add cookies from http session to web socket used by JupyterCodeExecutor by @timrbula in #3312
• Bump micromatch from 4.0.5 to 4.0.8 in /website by @dependabot in #3417
• Throttle API Calls by @WaelKarkoub in #2283
• Add kwargs to pass to docker container that runs the Jupyter Server by @rahulAira in #3028
• Remove api key validation by @jackgerrits in #3569
• Add init to agent_eval by @jackgerrits in #3568
• Add Couchbase VectorDB Support by @lokesh-couchbase in #3525
• WebSurfer Updated (Selenium, Playwright, and support for many filetypes) by @afourney in #1929
• Embedding function should always retur...

v0.2.35

What's Changed

• Removes Support For TransformChatHistory and CompressibleAgent by @WaelKarkoub in #3313
• Updated Program.cs for Autogen.BasicSample to give menu driven options by @cbelwal in #3346
• Remove dependency on RetrieveAssistantAgent for RetrieveChat by @thinkall in #3320
• Missing backticks breaking documentation in groupchat.last_speaker by @HenryKobin in #3357
• Update Mistral client class to support new Mistral v1.0.1 package by @marklysze in #3356
• adding a new page Function comparison between Python AutoGen and Auto… by @olaoluwasalami in #3367
• Update README.md to use camera-ready by @gagb in #3370
• Add OpenAI Gemini Example for VertexAI Notebook by @luxzoli in #3290
• [.NET] Add cache control to Anthropic Client by @DavidLuong98 in #3372
• Correct documentation: Replace NEVER with TERMINATE in human-in-the-loop.ipynb by @realmorrisliu in #3379
• update contact information on the repo and release package by @ekzhu in #3383

New Contributors

• @cbelwal made their first contribution in #3346
• @HenryKobin made their first contribution in #3357
• @realmorrisliu made their first contribution in #3379

Full Changelog: v0.2.34...v0.2.35

v0.2.34

Highlights

• Enhanced tool calling in Cohere
• Enhanced async support

What's Changed

• [CAP] Added a factory for runtime by @rajan-chari in #3216
• [Feature]: Add global silent param for ConversableAgent by @wenngong in #3244
• Fix Issue #2880: Document the usage of the AAD auth by @prithvi2226 in #2941
• [.Net] only add the last message to chat history in GroupChatExtension.SendAsync by @LittleLittleCloud in #3272
• [.Net] Remove Azure.AI.OpenAI from AutoGen.DotnetInteractive package by @LittleLittleCloud in #3274
• Correcting tool calling with Cohere by @jaygdesai in #3271
• Fix #3281 by @thinkall in #3282
• Fix autogen circular import error by @wenngong in #3276
• [.Net] Fix #3306 by @LittleLittleCloud in #3310
• [.Net] add DotnetInteractiveKernelBuilder to AutoGen.DotnetInteractive by @LittleLittleCloud in #3317
• [.Net] Add AutoGen.AzureAIInference by @LittleLittleCloud in #3332
• Support async nested chats by @heyitsaamir in #3309
• fix broken link to conversational chess example by @alu042 in #3327
• Add last_speaker to GroupChatManager by @thinkall in #3318
• Fix message history limiter for tool call by @GaoxiangLuo in #3178
• Fix to issue #3295 related to Anthropic bedrock by @makkzone in #3298
• [.Net][AutoGen.DotnetInteractive] add DotnetInteractiveStdioConnector by @LittleLittleCloud in #3337
• Add latest gpt-4o model: gpt-4o-2024-08-06 by @umermansoor in #3329
• Bump version by @qingyun-wu in #3343

New Contributors

• @jaygdesai made their first contribution in #3271
• @heyitsaamir made their first contribution in #3309
• @alu042 made their first contribution in #3327
• @GaoxiangLuo made their first contribution in #3178

Full Changelog: v0.2.33...v0.2.34