unique-orchestrator 1.13.5

No project description provided

Unique Orchestrator

Unique AI orchestrator agent to struct and direct subagents, tools and services

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[1.13.05] - 2026-02-26

• Add system prompt instruction for image rendering: use ![image](unique://content/[content_id]) when referring to tool-returned images

[1.13.04] - 2026-02-26

• Hide Responses API config from experimental section in UI (SkipJsonSchema); Code Interpreter is now configured via tools section only

[1.13.03] - 2026-01-23

• Changing parameter name from user_instruction to user_space_instruction

[1.13.02] - 2026-02-23

• Add flag-based cancellation support via CancellationWatcher

[1.13.01] - 2026-02-23

• Fix code interpreter tool toggle: auto-enable Responses API and populate default CodeInterpreterExtendedConfig when an enabled code_interpreter tool is present and model supports Responses API
• Fix _build_responses to filter tool_names by is_enabled so a disabled tool entry does not block the builder from adding an enabled instance

[1.13.00] - 2026-01-23

• Including functionality of user instruction in orchestrator

[1.12.02] - 2026-01-05

• Adjust feature flag

[1.12.01] - 2026-01-26

• Fix system prompt to enforce proper markdown list indentation for hierarchical formatting

[1.12.00] - 2026-01-23

• Use responses api version of LoopIterationHandler

[1.11.14] - 2026-01-21

• Add Qwen-specific loop iteration limits to reduce unnecessary agent runs

[1.11.13] - 2026-01-20

• Fix configuration for backward compatibility

[1.11.12] - 2026-01-16

• Add local CI testing commands via poethepoet (poe lint, poe test, poe ci-typecheck, etc.)

[1.11.11] - 2026-01-16

• Config cleanup of orchestrator

[1.11.10] - 2026-01-16

• Add unified type checking CI with basedpyright

[1.11.9] - 2026-01-15

• Add pytest-cov dev dependency for coverage testing

[1.11.8] - 2026-01-13

• Clean up dependencies and add deptry configuration for CI compliance

[1.11.7] - 2026-01-13

• Removing unused and retired configuration parameters max_review_steps, uploaded_content_config, tool_progress_reporter_config, force_checks_on_stream_response_references and thinking_steps_display

[1.11.6] - 2026-01-13

• Deprecate UniqueAIResponsesApi and use UniqueAI directly instead (type safety)

[1.11.5] - 2026-01-12

• Include feature flag to have message logs compatible with new ChatUI

[1.11.4] - 2025-12-29

• Fixes orchestrator tests and adds a ci test pipeline for it

[1.11.3] - 2025-12-29

• Fix system prompt with priority rule

[1.11.2] - 2025-12-18

• Improve bullet style for tool call message logs (for consistency)

[1.11.1] - 2025-12-11

• Improving support for forced tool call for Qwen models

[1.11.0] - 2025-12-11

• Add support for forced tool call for Qwen models

[1.10.0] - 2025-12-08

• Upgrading swot tool

## [1.9.0] - 2025-12-05

• Add support for planning before every loop iteration

## [1.8.2] - 2025-12-04

• Fix logging tool calls when tool takes over control

## [1.8.1] - 2025-12-03

• Logging tool calls when tool takes over control

## [1.8.0] - 2025-12-02

• Add option to upload code interpreter generated files to the chat.

## [1.7.19] - 2025-12-01

• Systemprompt formatting update

[1.7.18] - 2025-11-27

• Improvement of message log service to indicate number of tool calls per loop iteration

[1.7.17] - 2025-11-27

• Fixed an issue where the orchestrator failed when the number of tool calls exceeded the maximum allowed as defined in the configuration.
• Increased default value of max parallel tool calls from 5 to 15

## [1.7.16] - 2025-11-26

• Removing log of Deep Research tool call while keeping messages generated within the Deep Research call

## [1.7.15] - 2025-11-24

• Streamlining message log service for listing tools and being compatible with SWOT and DeepSearch

[1.7.14] - 2025-11-20

• Add message log service

[1.7.13] - 2025-11-20

• Fix bug of handling properly uploaded files that are expired

[1.7.12] - 2025-11-19

• Bump Swot tool

[1.7.11] - 2025-11-17

• Fix bug where forcing a tool still sends builtin tools to the LLM when using the responses api.

[1.7.10] - 2025-11-14

• Move pytest to dev dependencies

[1.7.9] - 2025-11-12

• Fix bug where Responses API config was not properly validated

[1.7.8] - 2025-11-11

• Better display of Responses API config in the UI

[1.7.7] - 2025-11-10

• Remove direct azure client config from responses api config
• Organize Responses API config better

[1.7.6] - 2025-11-05

• Update default system prompt (including user metadata section)

[1.7.5] - 2025-11-05

• Adding functionality to include user metadata into user/system prompts of the orchestrator

[1.7.4] - 2025-11-04

• Update and adapt to toolkit 1.23.0 (refactor sub agents implementation)

[1.7.3] - 2025-11-03

• Fixed an issue where new assistant messages were not properly generated during streaming outputs with tool calls; the orchestrator now correctly creates messages via _create_new_assistant_message_if_loop_response_contains_content when loop_response includes text and tool invocations.

[1.7.2] - 2025-11-03

• Add Swot tool to the orchestrator

[1.7.1] - 2025-10-30

• Fixing that system format info is only appended to system prompt if tool is called

[1.7.0] - 2025-10-30

• Add option to customize the display of tool progress statuses.
• Make follow-questions postprocessor run last to make sure the follow up questions are displayed last.

[1.6.1] - 2025-10-28

• Removing unused experimental config full_sources_serialize_dump in history_manager

[1.6.0] - 2025-10-27

• Add temporary config option sleep_time_before_update to avoid rendering issues with sub agent responses`

[1.5.2] - 2025-10-23

• Run evaluation and post processing in parallel

[1.5.1] - 2025-10-17

• revert behavior of unique ai upload and chat to

1. Add upload and chat tool to forced tools if there are tool choices
2. Simply force it if there are no tool choices.
3. Tool not available when no uploaded documents

[1.5.0] - 2025-10-16

• Make code interpreter configurable through spaces 2.0.

[1.4.3] - 2025-10-16

• Fix issue with openai base url

[1.4.2] - 2025-10-16

• Update debug info for better tool call tracking

[1.4.1] - 2025-10-16

• Temporarily make open ai env vars configurable

[1.4.0] - 2025-10-14

• Add responses api and code execution support.

[1.3.0] - 2025-10-14

• Re-organize sub-agents configuration for clarity.

[1.2.4] - 2025-10-14

• Let control taking tool itself set the message state to completed

[1.2.3] - 2025-10-13

• Fix bug where follow-up questions were being generated even if the number of questions is set to 0 in the config.

[1.2.2] - 2025-10-09

• update loading path of DEFAULT_GPT_4o from unique_toolkit

[1.2.1] - 2025-10-07

• upgrade to deep research 3.0.0

[1.2.0] - 2025-10-07

• Add sub agent response referencing.

[1.1.1] - 2025-10-03

• Adapt orchestrator to toolkit 1.8.0.

[1.1.0] - 2025-09-29

• Add ability to display sub agent's answers in main agent.
• Add ability to consolidate sub agent's assessment's in main agent.

[1.0.3] - 2025-09-29

• fix UniqueAI system prompt for not activated tools
• updated README

[1.0.2] - 2025-09-29

• updated deep-research to v2

[1.0.1] - 2025-09-18

• updated toolkit

[1.0.0] - 2025-09-18

• Bump toolkit version to allow for both patch and minor updates

[0.0.4] - 2025-09-17

• Updated to latest toolkit

[0.0.3] - 2025-09-16

• Cleaned configuration

[0.0.2] - 2025-09-15

• Resolve dependency bug

[0.0.1] - 2025-08-18

• Initial release of orchestrator.