unique_toolkit 0.8.29__tar.gz → 1.37.0__tar.gz

{unique_toolkit-0.8.29 → unique_toolkit-1.37.0}/CHANGELOG.md

@@ -4,6 +4,534 @@ All notable changes to this project will be documented in this file.
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/), 
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [1.37.0] - 2025-12-15
+- Adding a prompt appendix to enforce forced tool calls when using Qwen models
+
+## [1.36.0] - 2025-12-11
+- Add support for a sub agent tool system reminder when no references are present in the sub agent response.
+
+## [1.35.4] - 2025-12-10
+- Fix a potential stacktrace leak
+
+## [1.35.3] - 2025-12-10
+- Add option to ignore some options when calling the LLM for the planning step.
+
+## [1.35.2] - 2025-12-05
+- Increase speed of token reducer
+
+## [1.35.1] - 2025-12-05
+- Improve efficiency of token reducer if tool calls overshoot max token limit
+
+## [1.35.0] - 2025-12-04
+- Add `LoopIterationRunner` abstraction and support for planning before every loop iteration.
+
+## [1.34.1] - 2025-12-02
+- Update code interpreter tool instructions.
+
+## [1.34.0] - 2025-12-02
+- Add option to upload code interpreter generated files to the chat.
+
+## [1.33.3] - 2025-12-02
+- Fix serialization of ToolBuildConfig `configuration` field.
+
+## [1.33.2] - 2025-12-01
+- Upgrade numpy to >2.1.0 to ensure compatibility with langchain library
+
+## [1.33.1] - 2025-12-01
+- Add `data_extraction` to unique_toolkit
+
+## [1.33.0] - 2025-11-28
+- Add support for system reminders in sub agent responses.
+
+## [1.32.1] - 2025-12-01
+- Added documentation for the toolkit,some missing type hints and doc string fixes.
+
+## [1.32.0] - 2025-11-28
+- Add option to filter duplicate sub agent answers.
+
+## [1.31.2] - 2025-11-27
+- Added the function `filter_tool_calls_by_max_tool_calls_allowed` in `tool_manager` to limit the number of parallel tool calls permitted per loop iteration.
+
+## [1.31.1] - 2025-11-27
+- Various fixes to sub agent answers.
+
+## [1.31.0] - 2025-11-20
+- Adding model `litellm:anthropic-claude-opus-4-5` to `language_model/info.py`
+
+## [1.30.0] - 2025-11-26
+- Add option to only display parts of sub agent responses.
+
+## [1.29.4] - 2025-11-25
+- Add display name to openai builtin tools
+
+## [1.29.3] - 2025-11-24
+- Fix jinja utility helpers import
+
+## [1.29.2] - 2025-11-21
+- Add `jinja` utility helpers to `_common`
+
+## [1.29.1] - 2025-11-21
+- Add early return in `create_message_log_entry` if chat_service doesn't have assistant_message_id (relevant for agentic table)
+
+## [1.29.0] - 2025-11-21
+- Add option to force include references in sub agent responses even if unused by main agent response.
+
+## [1.28.9] - 2025-11-21
+- Remove `knolwedge_base_service` from DocXGeneratorService
+
+## [1.28.8] - 2025-11-20
+- Add query params to api operation
+- Add query params to endpoint builder
+
+## [1.28.7] - 2025-11-20
+- Adding Message Step Logger Class to the agentic tools.
+
+## [1.28.6] - 2025-11-20
+- Adding tests for message role filtering in chat functions
+
+## [1.28.5] - 2025-11-20
+- Including `expired_at` parameter in content schema
+- Including chatRole `SYSTEM` into chat schema
+
+## [1.28.4] - 2025-11-20
+- Bump tiktoken to 0.12.0
+
+## [1.28.3] - 2025-11-20
+- Add batch upload of files to knowledgebase
+- Add create folders utilities
+
+## [1.28.2] - 2025-11-20
+- Adding model `litellm:gemini-3-pro-preview` to `language_model/info.py`
+
+## [1.28.1] - 2025-11-19
+- Remove `chat_service` from DocXGeneratorService
+
+## [1.28.1] - 2025-11-19
+- Set review standards in pyright for toolkit
+- Refactor type check pipeline
+
+## [1.28.0] - 2025-11-19
+- Add option to interpret sub agent responses as content chunks.
+- Add option to specify a custom JSON schema for sub agent tool input.
+
+## [1.27.2] - 2025-11-19
+- Add test to token counting
+
+## [1.27.1] - 2025-11-18
+- Add missing `create_query_params_from_model` in experimental endoint_builder.py
+
+## [1.27.0] - 2025-11-18
+- Add `session_config` field to `ChatEventPayload` schema for chat session configuration support
+- Add `ingestion_state` field to `Content` model for tracking content ingestion status
+- Add `include_failed_content` parameter to content search functions in `KnowledgeBaseService`, `search_contents`, and `search_contents_async`
+- Experimental: 
+    - Fix HTTP GET requests in `build_request_requestor` to use query parameters (`params`) instead of JSON body
+    - `build_requestor` to properly pass kwargs to `build_request_requestor`
+
+## [1.26.2] - 2025-11-17
+- Adding tool format information for MCP tools
+
+## [1.26.1] - 2025-11-17
+- Fix bug where forcing a tool still sends builtin tools to the LLM when using the responses api.
+
+## [1.26.0] - 2025-11-17
+- Adding model `AZURE_GPT_51_2025_1113`, `AZURE_GPT_51_THINKING_2025_1113`, `AZURE_GPT_51_CHAT_2025_1113`, `AZURE_GPT_51_CODEX_2025_1113`,  `AZURE_GPT_51_CODEX_MINI_2025_1113` and `litellm:openai-gpt-51` and `litellm:openai-gpt-51-thinking` to `language_model/info.py`
+
+## [1.25.2] - 2025-11-12
+- Standardize paths in unique toolkit settings
+
+## [1.25.1] - 2025-11-12
+- Make pipeline steps reusable
+- Add pipeline checks for type errors
+- Add pipeline checks for coverage
+
+## [1.25.0] - 2025-11-10
+- Download files generated by code execution in parrallel.
+- Better display of files in the chat while loading.
+
+## [1.24.5] - 2025-11-10
+- Fix bug where images were not properly converted to responses api format.
+
+## [1.24.4] - 2025-11-07
+- Add `user_id` to language service for tracking
+- Track `user_id` in hallucination check
+
+## [1.24.3] - 2025-11-07
+- Adding litellm models `litellm:gemini-2-5-flash-lite`
+
+## [1.24.2] - 2025-11-06
+- Fix build_requestor typehints
+
+## [1.24.1] - 2025-11-06
+- Add IconChartBar Icon to ToolIcon
+
+## [1.24.0] - 2025-11-04
+- Introduce ability to include system reminders in tools to be appended when the response is included in the tool call history
+
+## [1.23.0] - 2025-11-04
+- Refactor sub agent tools implementation for clarity and testability.
+
+## [1.22.2] - 2025-11-03
+- Updated `unique_ai_how-it-works.md` and `plan_processing.md` to document how new assistant messages are generated when the orchestrator produces output text and triggers tool calls within the same loop iteration.
+
+## [1.22.1] - 2025-11-03
+- Add missing package required markdown-it-py 
+
+## [1.22.0] - 2025-10-31
+- Add `DocxGeneratorService` for generating Word documents from markdown with template support
+- Fix documentation for `update_message_execution`, `update_message_execution_async`, `update_assistant_message_execution`, and `update_assistant_message_execution_async` functions to correctly reflect that the `status` parameter is now optional
+
+## [1.21.2] - 2025-10-30
+- Fixing that system format info is only appended to system prompt if tool is called
+
+## [1.21.1] - 2025-10-30
+- Improve Spaces 2.0 display of tool progress reporter configuration.
+
+## [1.21.0] - 2025-10-30
+- Add option to customize the display of tool progress statuses.
+
+## [1.20.1] - 2025-10-30
+- Fix typing issues in `LanguageModelFunction`.
+
+## [1.20.0] - 2025-10-30
+- Fix bug where async tasks executed with `SafeTaskExecutor` did not log exceptions.
+- Add option to customize sub agent response display title.
+- Add option to display sub agent responses after the main agent response.
+- Add option to specify postprocessors to run before or after the others in the `PostprocessorManager`.
+
+## [1.19.3] - 2025-10-29
+- More documentation on advanced rendering
+
+## [1.19.2] - 2025-10-29
+- Removing unused tool specific `get_tool_call_result_for_loop_history` function
+- Removing unused experimental config `full_sources_serialize_dump` in `history_manager`
+
+## [1.19.1] - 2025-10-29
+- Make api operations more flexible
+- Make verification button text adaptable
+
+## [1.19.0] - 2025-10-28
+- Enable additional headers on openai and langchain client
+
+
+## [1.18.1] - 2025-10-28
+- Fix bug where sub agent references were not properly displayed in the main agent response when the sub agent response was hidden.
+
+## [1.18.0] - 2025-10-27
+- Temporary fix to rendering of sub agent responses.
+- Add config option `stop_condition` to `SubAgentToolConfig`
+- Add config option `tool_choices` to `SubAgentToolConfig`
+- Make sub agent evaluation use the name of the sub agent name if only one assessment is valid
+
+## [1.17.3] - 2025-10-27
+- Update Hallucination check citation regex parsing pattern
+
+## [1.17.2] - 2025-10-23
+- Adding model `AZURE_GPT_5_PRO_2025_1006` and `litellm:openai-gpt-5-pro` to `language_model/info.py`
+
+## [1.17.1] - 2025-10-23
+- Fix hallucination check input with all cited reference chunks.
+
+## [1.17.0] - 2025-10-22
+- Add more options to display sub agent answers in the chat.
+
+## [1.16.5] - 2025-10-16
+- Adding litellm models `litellm:anthropic-claude-haiku-4-5`
+
+## [1.16.4] - 2025-10-18
+- Fix bug with MCP tool parameters schema
+
+## [1.16.3] - 2025-10-18
+- Add logging of MCP tool schema and constructed parameters
+
+## [1.16.2] - 2025-10-16
+- Reduce operation dependency on path instead of full url
+
+## [1.16.1] - 2025-10-16
+- Update debug info for better tool call tracking
+
+## [1.16.0] - 2025-10-16
+- Add responses api support.
+- Add utilities for code execution.
+
+## [1.15.0] - 2025-10-15
+- Enable to distinguish between environment and modifiable payload parameters in human verification
+
+## [1.14.11] - 2025-10-13
+- Introduce testing guidelines for AI
+- Add AI tests to `tool_config` and `tool_factory.py`
+
+## [1.14.10] - 2025-10-13
+- Fix token counter if images in history
+
+## [1.14.9] - 2025-10-13
+- Fix removal of metadata
+
+## [1.14.8] - 2025-10-13
+- Use default tool icon on validation error
+
+## [1.14.7] - 2025-10-13
+- Update of token_limit parameters for Claude models
+
+## [1.14.6] - 2025-10-10
+- Fix circular import appearing in orchestrator
+
+## [1.14.5] - 2025-10-09
+- Move `DEFAULT_GPT_4o` constant from `_common/default_language_model.py` to `language_model/constants.py` and deprecate old import path
+
+## [1.14.4] - 2025-10-09
+- Small fixes when deleting content
+- Fixes in documentation
+
+## [1.14.3] - 2025-10-08
+- Reorganizes documentation 
+- All service interface towards a single folder
+- Add main imports to __init__
+
+## [1.14.2] - 2025-10-08
+- Add utilities for testing and fix external import issue
+
+## [1.14.1] - 2025-10-08
+- Add utilities for testing
+
+## [1.14.0] - 2025-10-07
+- Manipulate Metadata with knowledge base service
+
+## [1.13.0] - 2025-10-07
+- Delete contents with knowledge base service
+
+## [1.12.1] - 2025-10-07
+- Fix bug where failed evaluations did not show an error to the user.
+
+## [1.12.0] - 2026-10-07
+- Add the `OpenAIUserMessageBuilder` for complex user messages with images
+- More examples with documents/images on the chat
+
+## [1.11.4] - 2026-10-07
+- Make newer `MessageExecution` and `MessageLog` method keyword only
+
+## [1.11.3] - 2026-10-07
+- Move smart rules to content
+- Add to documentation
+
+## [1.11.2] - 2025-10-07
+- Fix for empty metadata filter at info retrieval
+
+## [1.11.1] - 2025-10-07
+- Fix bug where hallucination check was taking all of the chunks as input instead of only the referenced ones.
+
+## [1.11.0] - 2025-10-07
+- Add sub-agent response referencing.
+
+## [1.10.0] - 2025-10-07
+- Introduce future proof knowledgebase service decoupled from chat
+- Extend chat service to download contents in the chat
+- Update documentation
+
+## [1.9.1] - 2025-10-06
+- Switch default model used in evaluation service from `GPT-3.5-turbo (0125)` to `GPT-4o (1120)`
+
+
+## [1.9.0] - 2026-10-04
+- Define the RequestContext and add aihttp/httpx requestors
+
+## [1.8.1] - 2026-10-03
+- Fix bug where sub agent evaluation config variable `include_evaluation` did not include aliases for previous names.
+
+## [1.8.0] - 2026-10-03
+- Sub Agents now block when executing the same sub-agent multiple times with `reuse_chat` set to `True`.
+- Sub Agents tool, evaluation and post-processing refactored and tests added.
+
+## [1.7.0] - 2025-10-01
+- Add functionality to remove text in `get_user_visible_chat_history`
+
+## [1.6.0] - 2025-10-01
+- revert and simplify 1.5.0 
+
+## [1.5.1] - 2025-10-01
+- Fix filtering logic to raise a ConfigurationException if chat event filters are not specified
+
+## [1.5.0] - 2025-10-01
+- Allow history manager to fetch only ui visible text
+
+## [1.4.4] - 2025-09-30
+- Fix bugs with display of sub-agent answers and evaluations when multiple sub-agent from the same assistant run concurrently.
+
+## [1.4.3] - 2025-09-30
+- Fix bug with sub-agent post-processing reference numbers.
+
+## [1.4.2] - 2025-09-30
+- Adding litellm models `litellm:anthropic-claude-sonnet-4-5` and `litellm:anthropic-claude-opus-4-1`
+
+## [1.4.1] - 2025-09-30
+- Handle sub agent failed assessments better in sub agent evaluator.
+
+## [1.4.0] - 2025-09-29
+- Add ability to consolidate sub agent's assessments.
+
+## [1.3.3] - 2025-09-30
+- fix bug in exclusive tools not making them selectable
+
+## [1.3.2] - 2025-09-29
+- Auto-update unique settings on event arrival in SSE client
+
+## [1.3.1] - 2025-09-29
+- More documentation on referencing
+
+## [1.3.0] - 2025-09-28
+- Add utilitiy to enhance pydantic model with metadata
+- Add capability to collect this metadata with same hierarchy as nested pydantic models
+
+## [1.2.1] - 2025-09-28
+- Fix bug where camel case arguments were not properly validated.
+
+## [1.2.0] - 2025-09-24
+- Add ability to display sub agent responses in the chat.
+
+## [1.1.9] - 2025-09-24
+- Fix bug in `LanguageModelFunction` to extend support mistral tool calling.
+
+## [1.1.8] - 2025-09-23
+- Revert last to version 1.1.6
+
+## [1.1.7] - 2025-09-23
+- Introduce keyword only functions in services for better backward compatibility
+- Deprecatea functions that are using positional arguments in services 
+
+## [1.1.6] - 2025-09-23
+- Fix model_dump for `ToolBuildConfig`
+
+## [1.1.5] - 2025-09-23
+- Fix a circular import and add tests for `ToolBuildConfig`
+
+## [1.1.4] - 2025-09-23
+- First version human verification on api calls
+
+## [1.1.3] - 2025-09-23
+- Updated LMI JSON schema input type to include annotated string field with title
+
+## [1.1.2] - 2025-09-22
+- Fixed bug tool selection for exclusive tools
+
+## [1.1.1] - 2025-09-18
+- Fixed bug on tool config added icon name
+
+## [1.1.0] - 2025-09-18
+- Enable chat event filtering in SSE event generator via env variables
+
+## [1.0.0] - 2025-09-18
+- Bump toolkit version to allow for both patch and minor updates
+
+## [0.9.1] - 2025-09-17
+- update to python 3.12 due to security 
+
+## [0.9.0] - 2025-09-14
+- Moved agentic code into the `agentic` folder. This breaks imports of
+  - `debug_info_amanager`
+  - `evals`
+  - `history_manager`
+  - `post_processor`
+  - `reference-manager`
+  - `short_term_memory_manager`
+  - `thinking_manager`
+  - `tools`
+
+## [0.8.57] - 2025-09-14
+- Added more utils to commons
+
+## [0.8.56] - 2025-09-12
+- Fixed token counter in utils
+
+## [0.8.55] - 2025-09-10
+- Update documentation with agentic managers
+
+## [0.8.54] - 2025-09-10
+- HistoryManager: compute source numbering offset from prior serialized tool messages using `load_sources_from_string`
+- LoopTokenReducer: serialize reduced tool messages as JSON arrays to keep offsets parsable
+
+## [0.8.53] - 2025-09-09
+- Add support for skip ingestion for only excel files.
+
+## [0.8.52] - 2025-09-06
+- Fix import error in token counting
+
+## [0.8.51] - 2025-09-06
+- Update token counter to latest version of monorepo.
+
+## [0.8.50] - 2025-09-08
+- Minor fix in documentation
+- Updated examples
+
+## [0.8.49] - 2025-09-05
+- Fixed token reducer now has a safety margin of 10% less did not work.
+
+## [0.8.48] - 2025-09-05
+- Add documentation on language models to markdown
+
+## [0.8.47] - 2025-09-05
+- Removed old code
+- Fixed small bugs in history manager & set the hallucination to use gpt4o as default.
+
+## [0.8.46] - 2025-09-04
+- Bugfix for hostname identification inside Unique cluster in `unique_settings.py`
+
+## [0.8.45] - 2025-09-04
+- Introduce handoff capability to tools. with the `takes_control()` function.
+
+## [0.8.44] - 2025-09-03
+- Refine `EndpointClass` and create `EndpointRequestor`
+
+## [0.8.43] - 2025-09-03
+- Add alias for `UniqueSettings` api base `API_BASE`
+
+## [0.8.42] - 2025-09-02
+- updated schema of `chunk_relevancy_sorter`
+
+## [0.8.41] - 2025-09-02
+- Make A2A tool auto register with tool factory
+
+## [0.8.40] - 2025-09-02
+- Add frontend compatible type for pydantic BaseModel types in pydantic BaseModels
+
+## [0.8.39] - 2025-09-02
+- include `get_async_openai_client`
+
+## [0.8.38] - 2025-09-01
+- Sanitize documentation
+
+## [0.8.37] - 2025-09-01
+- Adapt defaults and json-schema in `LanguageModelInfo`
+
+## [0.8.36] - 2025-09-01
+- Added dependency `Pillow` and `Platformsdir`
+
+## [0.8.35] - 2025-09-01
+- Initial toolkit documentation (WIP)
+
+## [0.8.34] - 2025-09-01
+- Automatic initializations of services and event generator
+
+## [0.8.33] - 2025-08-31
+- fixed tool for `web_search`
+
+## [0.8.32] - 2025-08-30
+- moved over general packages for `web_search`
+
+## [0.8.31] - 2025-08-29
+- Add various openai models to supported model list
+  - o1
+  - o3
+  - o3-deep-research
+  - o3-pro
+  - o4-mini
+  - o4-mini-deep-research
+  - gpt-4-1-mini
+  - gpt-4-1-nano
+
+## [0.8.30] - 2025-08-28
+- Added A2A manager
+
 ## [0.8.29] - 2025-08-27
 - Include `MessageExecution` and `MessageLog` in toolkit 
 
@@ -16,6 +544,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [0.8.26] - 2025-08-27
 - Optimized MCP manager
 
+## [0.8.26] - 2025-08-27
+- Optimized MCP manager
+
 ## [0.8.25] - 2025-08-27
 - Load environment variables automatically from plattform dirs or environment
 - General Endpoint definition utility
@@ -521,4 +1052,4 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   - `utils.py` with utility functions for parsing language model output.
 
 ## [0.0.2] - 2024-07-10
-- Initial release of `unique_toolkit`.
+- Initial release of `unique_toolkit`.