v1.11.0

What's Changed

• Bump dev version: v1.11.0.dev0 by @albertvillanova in #890
• Add Chinese translation & fix zh doc build process by @muscaestar in #893
• Fix chained assignment in local executor by @sysradium in #843
• Check dangerous modules instead of dangerous patterns by @albertvillanova in #877
• Fix typo in building good agents tutorial by @whistler in #903
• Add tests and refactor CLI by @albertvillanova in #892
• Bump mcpadapt >= 0.0.15 by @albertvillanova in #895
• Update docs about changing the system prompt by @albertvillanova in #900
• Support transformers 4.49.0 by @albertvillanova in #898
• Fix typo in list of dangerous functions by @albertvillanova in #909
• Improve evaluate_subscript by @albertvillanova in #916
• Enforce security check in all evaluations by @albertvillanova in #906
• Improve performance of test_while infinite loop by @albertvillanova in #907
• Test all dangerous functions and fix check of os.system by @albertvillanova in #911
• Fix Tool docstring, and model init exception by @zhanluxianshen in #927
• Remove default model_id from models by @albertvillanova in #905
• Improve docstring of parse_code_blobs by @albertvillanova in #876
• Test vulnerability for dangerous modules by @albertvillanova in #929
• Fix broken link to benchmarking code in README by @amaarora in #933
• Fix Gradio UI for unnamed agents by @sysradium in #812
• Add mlx-lm to all extra by @albertvillanova in #961
• Fix Gradio UI for None name by @albertvillanova in #956
• Introduce counter class by @Bilokin in #891
• Fix AzureOpenAIServerModel to not call openai.OpenAI by @albertvillanova in #966
• Fix CI usage of uv by @albertvillanova in #970
• Adds VLLMModel by @NathanHB in #337
• Support passing flatten_messages_as_text to OpenAIServerModel and test by @albertvillanova in #959
• Fix tool function implementation by @albertvillanova in #943
• Fix broken example of using MCP Server by @njbrake in #936
• Support passing any kwargs to E2BExecutor Sandbox by @aymeric-roucher in #951
• Fix VLLMModel example in models docs by @zhanluxianshen in #976
• Fix OpenAIServerModel client by @durd07 in #975
• Forbid all modules by default except whitelist authorized_imports by @albertvillanova in #935
• Forbid access to all dunder attributes by default by @albertvillanova in #979
• Raise agent generation errors by @aymeric-roucher in #948
• Update Open Deep Research's README.md by @keetrap in #763
• Make sure to log thoughts if verbosity_level is set to high by @aymeric-roucher in #964
• Get correct source code for functions defined in local executor. by @aymeric-roucher in #973

New Contributors

• @muscaestar made their first contribution in #893
• @whistler made their first contribution in #903
• @zhanluxianshen made their first contribution in #927
• @amaarora made their first contribution in #933
• @NathanHB made their first contribution in #337
• @njbrake made their first contribution in #936
• @durd07 made their first contribution in #975

Full Changelog: v1.10.0...v1.11.0

v1.10.0

New features 🚀🚀

• Add Docker executor by @aymeric-roucher in #733
  • This means you can now initialize agents with executor="docker" to run code in a Docker sandbox! Docs coming soon.
• Safer local Python interpreter
  • Make local Python interpreter safer by checking if returns builtins by @albertvillanova in #848
  • Make local Python interpreter safer by checking if returns dangerous modules by @albertvillanova in #861
  • Make local Python interpreter safer by adding builtins to dangerous modules by @albertvillanova in #869
• Add support for serper search by @aymeric-roucher in #698
• Support running a demo of Open DeepResearch by @albertvillanova in #806
• Enable Colab notebook version of each doc page (by renaming doc file extension from .md to .mdx) by @albertvillanova in #691

Fixes and improvements 🔧

• Bump dev version: v1.10.0.dev0 by @albertvillanova in #649
• Improve missing toolcall exception by @GrahamTheCoder in #614
• refactor: simplify file type checking from MIME to extension by @kingdomad in #342
• Bugfix: Groq via LiteLLM 🚅 by @JGalego in #605
• Add docstring args for MultiStepAgent.from_folder by @albertvillanova in #654
• Remove non-existent subfolder arg in MultiStepAgent.from_hub by @albertvillanova in #653
• Minor fix save agents by @keetrap in #645
• Align step numbers in monitoring logging by @spemed in #667
• Remove deprecated resume_download arg in from_hub by @albertvillanova in #670
• Pass notebook_folder to build notebook docs by @albertvillanova in #671
• Update the prompts for delegating subtasks to agents in prompt files by @spemed in #678
• Documentation: manage agent's memory by @aymeric-roucher in #675
• Move contributing details into contributing file by @GrahamTheCoder in #651
• Fix typo in error message by @samee99 in #687
• Refactor evaluate ast to improve readability by @CalOmnie in #625
• Update docs in zh/examples/multiagents.md by @spemed in #688
• Make dangerous_patterns a module variable by @CalOmnie in #505
• Pin transformers < 4.49.0 to fix TypeError: LlavaProcessor: got multiple values for keyword argument 'images' by @albertvillanova in #693
• Respect formatting in ast.FormattedValue for local_python_executor by @LorienMan in #660
• Simplify instrumentation for phoenix by @aymeric-roucher in #697
• Test validate_tool_attributes by @albertvillanova in #696
• Remove translation tool from README by @keetrap in #705
• Fix test docs by @albertvillanova in #701
• Fix SpeechToTextTool by @albertvillanova in #706
• Fixed typo in Telemetry docs by @sergiopaniego in #710
• Fix validate_tool_attributes for non-literal defaults by @albertvillanova in #711
• Ignore parameter types in tool docstrings by @carlsberg03 in #724
• Refactor MultiStepAgent and improve some tests by @colesmcintosh in #598
• Fix duplicate managed agent name and its tool name by @keetrap in #721
• Update docs on how to change prompts by @albertvillanova in #725
• Remove transformers tasks from load_tool docstring by @albertvillanova in #726
• Remove article section from launch_gradio_demo Interface by @albertvillanova in #728
• Update logo in Readme by @aymeric-roucher in #730
• Support passing additional parameters for openai client initialization by @touseefahmed96 in #388
• Add api-base and api-key arguments to CLI by @keetrap in #734
• Fix CI test error: DuckDuckGoSearchException: 202 Ratelimit by @albertvillanova in #739
• Fix typo in Makefile by @albertvillanova in #741
• Add disclaimer in benchmark to request access for gated dataset by @albertvillanova in #744
• Define test RUN_ALL requirement marker in one file by @albertvillanova in #743
• Rename managed agent in multiagents docs to avoid confusing the LLM by @albertvillanova in #766
• Update documentation on sandboxes by @aymeric-roucher in #747
• Add audioop-lts as Python>=3.13 Open DeepResearch requirement for pydub by @albertvillanova in #749
• Create docker extra installation dependency by @albertvillanova in #764
• Fix bracket logging in rich by @aymeric-roucher in #753
• Enable custom final answer in CodeAgent by @aymeric-roucher in #769
• Fix TypeError raised by launch_gradio_demo with PipelineTool by @albertvillanova in #732
• Use docker extra in Secure code execution docs by @albertvillanova in #780
• Support custom final_answer in agent init by @albertvillanova in #783
• Fix LiteLLMModel API key usage in CLI by @asselindebeauville in #788
• Secure code execution wrong title in docs TOC by @maeste in #771
• Update Open DeepResearch to install smolagents from local path by @albertvillanova in #786
• Support importing scipy and sklearn without AttributeError: module 'scipy.sparse._coo' has no attribute 'upcast' by @albertvillanova in #790
• Fix and test validate_tools_and_managed_agents by @albertvillanova in #731
• Update README with mention to Docker sandboxed execution by @touseefahmed96 in #803
• Added sidebar layout and session state management for agent by @yvrjsharma in #797
• Fix CI quality in examples/gradio_ui by @albertvillanova in #822
• Improve dependency loading of Open DeepResearch visual_qa by @albertvillanova in #821
• Improve setup instruction for Open DeepResearch by @albertvillanova in #823
• Support other models than o1 in Open Deep Research by @albertvillanova in #824
• Fix final_answer_pattern matching in remote executors by @albertvillanova in #826
• Fix and refactor tests of Docker executors by @albertvillanova in #827
• Remove unnecessary self argument in GoogleSearchTool super() call by @gael1130 in #828
• Fix call from dict with LocalPythonExecutor by @albertvillanova in #837
• Minor fix of import in VisitWebpageTool by @keetrap in #840
• No need of these Files by @touseefahmed96 in #798
• fix: correcting tool_call role by @zndqs in #779
• Fix translation guide as it points to invalid location by @FanwangM in #794
• Fix load_model in webagent CLI by @albertvillanova in #855
• Fix message for observations without tool_calls by @albertvillanova in #857
• Fix CI test_action_step_to_messages by @albertvillanova in #868
• Fix LiteLLMModel docstring by @albertvillanova in #885
• Fix bug where operations count was not reinitialized at each step by @aymeric-roucher in #866
• Fix CLI webagent by @albertvillanova in https://github.com/huggingface/smolage...

v1.9.2

Bug Fixes 🐛

• Fix litellm flatten_messages_as_text detection by @aymeric-roucher in ef60315
• Fix litellm flatten_messages_as_text detection by @aymeric-roucher in #659

Full Changelog: v1.9.1...v1.9.2

v1.9.1

Bug Fixes 🐛

• fix tool code generation in case of multiline descriptions by @sysradium in #613
• Fix prompt for toolcalling agent and slightly update doc by @aymeric-roucher in #650

Full Changelog: v1.9.0...v1.9.1

v1.9.0

What's Changed

• Bump dev version: v1.9.0.dev0 by @albertvillanova in #544
• Small update to 'Introduction to Agents' docs format by @sergiopaniego in #548
• Doc grammar fix by @juanqui in #556
• Minor grammar correction by @doomspork in #528
• Fix multiagents docs by @derekalia in #549
• Fix managed agent example by @keetrap in #562
• Add authorized_imports in importFrom errors by @CalOmnie in #506
• chore: fix some typos by @zoupingshi in #557
• Minor fix in the guilded_tour docs by @touseefahmed96 in #578
• Add small readme to open Deep Research by @aymeric-roucher in #581
• Fixed Readme that is missing run.py call by @jsmidt in #585
• Fix 429 Client Error: Too Many Requests in CI test by @albertvillanova in #599
• Remove unused import in guided_tour docs by @albertvillanova in #600
• Test evaluate_delete by @albertvillanova in #519
• Move check_module_authorized out of import_module for use in get_safe_module by @CalOmnie in #507
• Test agent call with provide_run_summary by @albertvillanova in #513
• Create PromptTemplates typed dict by @albertvillanova in #547
• Refactor and test final answer prompts by @albertvillanova in #595
• Make question arg required in Open DeepResearch example by @albertvillanova in #617
• Remove unused api-base arg from Open DeepResearch example by @albertvillanova in #616
• corrected typo in README.md by @blakkd in #609
• MLX model support by @g-eoj in #300
• Add support for non-bool comparison operators. by @kc9zyz in #612
• docs: add Langfuse OpenTelemetry guide by @jannikmaierhoefer in #601
• Remove --prompt argument of instance instruction in README.md by @leeivan1007 in #632
• Fix CI quality check by removing trailing whitespace by @albertvillanova in #628
• Change math.pow -> pow by @Bilokin in #624
• Move plan user prompt to YAML and test text of plan prompts by @albertvillanova in #591
• Fix installation instructions in Open-DeepResearch README by @albertvillanova in #633
• Remove ManagedAgent from doc by @aymeric-roucher in #563
• Refactor operations count state setting by @albertvillanova in #631
• LiteLLMModel - detect message flatenning based on model information by @sysradium in #553
• Contribute to the documentation by @seanxuu in #630
• Adding default parameter for max_new_tokens in TransformersModel by @matfrei in #604
• Test evaluate_condition by @albertvillanova in #634
• Allow Gradio share parameter passthrough by @sysradium in #490
• Fix issue #635. Corrected installation instructions for open_deep_research by @nishaddeokar in #636
• Share full agents by @aymeric-roucher in #533
• Fix PlanningPromptTemplate and its docstring by @albertvillanova in #648
• Fix evaluate_condition for non-bool result by @albertvillanova in #638
• Removed reference to deprecated single_step in documentation by @crumbly-blue-cheese in #608
• Update README.md by @aymeric-roucher in #642
• Fix reference to Transformers in docs by @sergiopaniego in #643
• Fix for MLX Model when stop sequence is followed by any chars by @g-eoj in #623
• Allow for last_input_token_count=None by @Strings-RH in #644

New Contributors

• @juanqui made their first contribution in #556
• @doomspork made their first contribution in #528
• @CalOmnie made their first contribution in #506
• @zoupingshi made their first contribution in #557
• @NeuroWhAI made their first contribution in #576
• @jsmidt made their first contribution in #585
• @blakkd made their first contribution in #609
• @g-eoj made their first contribution in #300
• @kc9zyz made their first contribution in #612
• @jannikmaierhoefer made their first contribution in #601
• @leeivan1007 made their first contribution in #632
• @Bilokin made their first contribution in #624
• @seanxuu made their first contribution in #630
• @matfrei made their first contribution in #604
• @nishaddeokar made their first contribution in #636
• @crumbly-blue-cheese made their first contribution in #608
• @Strings-RH made their first contribution in #644

Full Changelog: v1.8.1...v1.9.0

v1.8.1

Bug Fixes 🐛

• Restore missing user prompt for initial facts by @NeuroWhAI in #576
• Bugfix: problems with the prompt refactoring of the planning step by @IlyaGusev in #564
• Do not pass system prompt to update plan and test plan prompts by @albertvillanova in #586

Full Changelog: v1.8.0...v1.8.1

v1.8.0

New features ⭐️

• We added visualization method to display the agent' structure as a tree! 🌳 by @aymeric-roucher in #470

• We vastly simplified managed agents in #484
  • We've removed the ManagedAgents class: directly provide name and description attributes to your agent to make it work as a managed agent ! Check an example here.
• We released Open Deep Research in #317!

Fixes 🔧

• Bump dev version: v1.8.0.dev0 by @albertvillanova in #451
• Add ToC and CLI docs by @merveenoyan in #455
• Fix link to Helium repository in README.md by @dross20 in #471
• Remove default max_tokens for all models by @albertvillanova in #475
• Improve Readme: shorter intro, VLMs, MCP, etc by @aymeric-roucher in #459
• Fix SpeechToTextTool new instance by @albertvillanova in #478
• Allow flatten_messages_as_text to be pushed via LiteLLMModel to fix Ollama integration by @sysradium in #406
• Fixed demo typo in README.md by @hanteed in #481
• Added HfApiModel import to web browser example by @sergiopaniego in #461
• Minor Fix tool push_to_hub by @keetrap in #460
• Fix link in README.md by @vanrohan in #496
• Fix undefined step_number in single_step run by @taha-yassine in #495
• Bump litellm to 1.60.2 (fixes #34) by @taha-yassine in #488
• Initialize step_number attribute by @albertvillanova in #497
• Suppress terminal logging in CI tests by @albertvillanova in #504
• Fix link in README.md by @Ryanb58 in #492
• Test PrintContainer by @albertvillanova in #515
• Add missing default tools to reference docs by @touseefahmed96 in #516
• Refactor prompts by @aymeric-roucher in #502
• Fix Typo by @rehanzo in #529
• Test HfApiModel call with custom_role_conversions by @albertvillanova in #517
• Set MAX_WHILE_ITERATIONS as module variable by @albertvillanova in #520
• add self.logger = logger for e2b_executor by @femto in #537
• Create simpler example open-deep-research by @aymeric-roucher in #534
• Support disabling logging to terminal by @albertvillanova in #512
• Fix installation with data files by @albertvillanova in #536
• Pin openinference-instrumentation-smolagents >= 0.1.4 by @albertvillanova in #535
• Delete prompts_path argument and use prompt_templates by @albertvillanova in #541
• Fix deprecation warning from importlib.resources.read_text by @albertvillanova in #543
• Delete non-exisiting accelerate extra by @albertvillanova in #539
• Fix some doc examples by @aymeric-roucher in #542
• Test E2B Executor by @albertvillanova in #540

New Contributors

• @dross20 made their first contribution in #471
• @sysradium made their first contribution in #406
• @hanteed made their first contribution in #481
• @vanrohan made their first contribution in #496
• @taha-yassine made their first contribution in #495
• @Ryanb58 made their first contribution in #492
• @rehanzo made their first contribution in #529
• @femto made their first contribution in #537

Full Changelog: v1.7.0...v1.8.0

v1.7.0

Main changes

Added commands to directly run agents from CLI: smolagent and webagent ✨

• Make web browser agent example a CLI script by @merveenoyan in #416
• Add CLI for smolagents by @aymeric-roucher in #431
  • Now you can run smolagent "Your task!" from CLI to directly call a smolagent. Try it! webagent "Find me a cheap train from Paris to Torino before Thursday" or smolagent "Who's the current director of La French Tech?"

Gave a memory to agents! 🧠

• Untangling Logging by @clefourrier in #316
• Test memory by @albertvillanova in #428
  • This added a specific memory attribute to the agent!
  • We used this change to introduce nice functions like an agent.replay() function that replays the last run (based on the memories stored, so no further LLM calls)

Other big improvements 💪

• Improve code execution logs in case of error by showing print outputs by @aymeric-roucher in #446

  • This makes sure that code execution outputs are stored to memory even when an error was raised later on. It should slightly improve the performance of CodeAgents

• Update benchmark with Hub datasets by @albertvillanova in #412

  • This adds storing of benchmarking outputs directly to the Hub. Moving forward, we'll make benchmarking even easier!

Fixes and small improvements

• Bump dev version: v1.7.0.dev0 by @albertvillanova in #390
• Fix return type of generator function in Agent._run() by @abidlabs in #409
• Update minimum Gradio version by @abidlabs in #408
• Support third-party Inference providers in HfApiModel by @julien-c in #422
• Minor fix for agent_from_any_llm.py example by @RolandJAAI in #423
• Fix vlm_web_browser.py example by @abidlabs in #410
• Fix regressions, refactor logging and improve replay function by @aymeric-roucher in #419
• Fix installation instruction for GradioUI in error message by @jank in #418
• Fix import from deleted logger module by @albertvillanova in #433
• Add web search tool and make prompt positional arg by @aymeric-roucher in #435
• Remove explicit model arg in HfApiModel - which wouldn't run without Pro by @sanjeed5 in #373
• Fix MultiStepAgent.planning_step message content by @albertvillanova in #437
• Fix and test MemoryStep by @albertvillanova in #432
• Fix get safe module by @antoinejeannot in #405
• Remove duplicate final_memory_step creation by @albertvillanova in #444
• Remove unused ChatMessageToolCall raw argument by @albertvillanova in #445
• Fix TransformersModel docstring by @August-murr in #447
• Test get_clean_message_list by @albertvillanova in #448
• Add web browser example to doc by @aymeric-roucher in #439
• Fix minor issue in e2b by @keetrap in #436

New Contributors

• @abidlabs made their first contribution in #409
• @julien-c made their first contribution in #422
• @August-murr made their first contribution in #447

Full Changelog: v1.6.0...v1.7.0

v1.6.0

What's Changed

• Bump version to 1.6.0.dev by @albertvillanova in #348
• Use transformers as optional dependency by @albertvillanova in #360
• Add pip install datasets to RAG example by @sanjeed5 in #371
• Update issue templates by @aymeric-roucher in #377
• TransformersModel auto-detects VLMs by @aymeric-roucher in #378
• Separate tree for Model docs is added by @touseefahmed96 in #382
• add kwargs to gradio launch by @tocab in #367
• Minor Fix in the example docs by @touseefahmed96 in #383
• Fix Extra planning by @bvantuan in #362
• Gradio chatbot: step duration, number, token count, support nested thoughts by @yvrjsharma in #384
  • Gradio Chatbot looks much nicer than before! See below:

• Add Chinese translations for text-to-sql example by @FanwangM in #366
• Add Chinese translations for RAG by @FanwangM in #363
• Qwen-2VL -> Qwen2-VL model name typo fixed by @sergiopaniego in #349
• chore: Fix Typo by Calling is_torch_available() in AgentAudio Class by @hironow in #379
• Additional parameters for openai client by @touseefahmed96 in #385
• DuckDuckGoSearchTool: add ddgs_kwargs parameter to constructor by @onukura in #372
• Fix final_answer issue in e2b_executor by @keetrap in #319
• Docs: Text to sql example - minor naming / dedupe by @GrahamTheCoder in #376
• Test import without extras by @albertvillanova in #370
• Add tool saving test by @aymeric-roucher in #389
• Fix blocking of os in authorized imports by @aymeric-roucher in #386
• Add test for DDGS Tool kwargs by @aymeric-roucher in #387

New Contributors

• @tocab made their first contribution in #367
• @bvantuan made their first contribution in #362
• @sergiopaniego made their first contribution in #349
• @hironow made their first contribution in #379
• @onukura made their first contribution in #372
• @GrahamTheCoder made their first contribution in #376

Full Changelog: v1.5.1...v1.6.0

v1.5.1

Bug Fixes 🐛

• Use transformers as optional dependency by @albertvillanova in #360

Full Changelog: v1.5.0...v1.5.1