Release : 0.4.61 Adding JSON structure output, Streaming, Nested steps, Updates, New dependacies...

[Zochory]

User description

This pull request includes :

Implemented multi-step visualization and word-by-word streaming in the Chainlit app:

Multi-step Visualization:

• Hierarchical step tracking with proper nesting
• Clear step types and metadata
• Progress tracking through TaskList
• Word-by-Word Streaming:
• Smooth text output with 0.01 second delay between words
• Proper word spacing and formatting
• Integrated with Chainlit's streaming capabilities
• Maintains message attribution and context
• Combined Features:
• Each streamed message is properly tracked in its step
• Step metadata shows full input/output
• Maintains all existing functionality (task tracking, error handling, etc.)

The implementation now provides both a clear visualization of the reasoning process through steps and a smooth, readable output through word-by-word streaming. This makes the agent interactions more dynamic and easier to follow.

Several updates to the documentation, configuration files, and dependencies of the agentic-fleet project. The most important changes include updates to the README.md file to improve clarity and add new sections, adjustments to dependencies in pyproject.toml, and the removal of an example script.

Documentation updates:

• README.md: Added a new "Core Components" section describing the roles of specialized agents, updated the "Key Features" section, and included a new "System Architecture" diagram and "Error Handling" section. [1] [2] [3] [4] [5]

Dependency updates:

• pyproject.toml: Updated project version to 0.4.61, added new dependencies (autogen-ext[langchain], seaborn, scikit-learn, ipykernel, authlib, starlette, semantic-kernel, tiktoken), and updated existing dependencies (chainlit to 2.0.6). [1] [2] [3] [4] [5]

Configuration file updates:

• .chainlit/config.toml: Updated the generated_by field to 2.0.6.

Example script removal:

• examples/quickstart/web_example.py: Removed the example script demonstrating AgenticFleet with Chainlit.

New experimental configuration:

• src/agentic_fleet/_experimental_config/planner/planner_memory.yaml: Added a new configuration file for summarizing error resolutions and preferences from chat history.…pendencies and enhance README documentation

---

PR Type

Enhancement, Documentation, Tests

---

Description

• Introduced multi-step visualization and word-by-word streaming for enhanced interaction.

  • Added stream_text function for streaming text word-by-word.
  • Enhanced response processing with step visualization using cl.Step.

• Added experimental configuration files for planner behavior and error handling.

  • Included planner_memory.yaml and planner_prompt.yaml for structured planning and error resolution.

• Updated documentation with detailed system architecture and core components.

  • Expanded README with new sections on error handling, architecture, and agent roles.

• Updated dependencies and configuration for improved functionality and compatibility.

  • Upgraded chainlit to version 2.0.6 and added new dependencies like seaborn and semantic-kernel.

---

Changes walkthrough 📝

	Relevant files
Miscellaneous	web_example.py Removed deprecated web example script                                        examples/quickstart/web_example.py Removed the deprecated web example script. Simplified project structure by eliminating unused examples. +0/-99
Enhancement	app.py Enhanced response processing and streaming                              src/agentic_fleet/app.py Added stream_text function for word-by-word text streaming. Enhanced response and message processing with step visualization. Improved task handling with structured steps and error handling. +182/-106 planner_memory.yaml Added planner memory configuration                                              src/agentic_fleet/_experimental_config/planner/planner_memory.yaml Added configuration for summarizing error resolutions and preferences. Defined structure for analyzing chat history errors. +27/-0    planner_prompt.yaml Added planner prompt configuration                                              src/agentic_fleet/_experimental_config/planner/planner_prompt.yaml Added detailed instructions for planner behavior and planning process. Defined JSON schema for planner responses. +192/-0
Configuration changes	config.toml Updated Chainlit configuration version                                      .chainlit/config.toml Updated generated_by version to 2.0.6. +1/-1
Documentation	README.md Expanded documentation with architecture and features        README.md Added detailed system architecture and core components. Expanded sections on error handling and agent roles. Updated setup instructions and added a star history chart. +77/-19
Dependencies	requirements.txt Removed outdated requirements file                                              examples/quickstart/requirements.txt Removed deprecated requirements file. +0/-3      pyproject.toml Updated project version and dependencies                                  pyproject.toml Updated project version to 0.4.61. Added new dependencies like seaborn, semantic-kernel, and tiktoken. Upgraded chainlit to version 2.0.6. +13/-4

---

Need help?

Type /help how to ... in the comments thread for any questions about Qodo Merge usage.

Check out the documentation for more information.

[qodo-merge-pro-for-open-source]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 4 🔵🔵🔵🔵⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Error Handling The new streaming functionality lacks proper error handling for network interruptions and timeouts during word-by-word streaming. This could lead to incomplete or stuck messages. async def stream_text(text: str) -> AsyncGenerator[str, None]: """Stream text content word by word. Args: text: Text to stream Yields: Each word of the text with a delay """ words = text.split() for i, word in enumerate(words): await asyncio.sleep(STREAM_DELAY) yield word + (" " if i < len(words) - 1 else "") Resource Management The nested steps implementation may lead to resource leaks if steps are not properly closed in case of exceptions. Consider using try-finally blocks. async with cl.Step(name="Response Processing", type="process") as main_step: main_step.input = str(response) # Handle TaskResult objects if isinstance(response, TaskResult): async with cl.Step(name="Task Execution", type="task") as task_step: task_step.input = getattr(response, 'task', 'Task execution') for msg in response.messages: await process_message(msg, collected_responses) if response.stop_reason: task_step.output = f"Task stopped: {response.stop_reason}" await cl.Message( content=f"🛑 {task_step.output}", author="System" ).send() # Handle TextMessage objects directly elif isinstance(response, TextMessage): async with cl.Step(name=f"Agent: {response.source}", type="message") as msg_step: msg_step.input = response.content await process_message(response, collected_responses) # Handle chat messages elif hasattr(response, 'chat_message'): async with cl.Step(name="Chat Message", type="message") as chat_step: chat_step.input = str(response.chat_message) await process_message(response.chat_message, collected_responses) # Handle inner thoughts and reasoning elif hasattr(response, 'inner_monologue'): async with cl.Step(name="Inner Thought", type="reasoning") as thought_step: thought_step.input = response.inner_monologue await cl.Message( content=f"💭 Inner thought: {response.inner_monologue}", author="System", indent=1 ).send() thought_step.output = "Processed inner thought" # Handle function calls elif hasattr(response, 'function_call'): async with cl.Step(name="Function Call", type="function") as func_step: func_step.input = str(response.function_call) await cl.Message( content=f"🛠️ Function call: {response.function_call}", author="System", indent=1 ).send() func_step.output = "Function call processed" # Handle multimodal messages (images, etc.) elif isinstance(response, (list, tuple)): async with cl.Step(name="Multimodal Content", type="media") as media_step: media_step.input = "Processing multimodal content" await _process_multimodal_message(response) media_step.output = "Multimodal content processed" # Handle any other type of response else: async with cl.Step(name="Generic Response", type="other") as generic_step: content = str(response) generic_step.input = content await cl.Message(content=content, author="System").send() collected_responses.append(content) generic_step.output = "Response processed" Performance The fixed STREAM_DELAY of 0.01 seconds could cause performance issues with large text blocks. Consider making this configurable or adaptive based on content length. STREAM_DELAY = 0.01

[qodo-merge-pro-for-open-source]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Score
Possible issue	Prevent infinite step nesting The process_response function should handle potential infinite recursion by adding a maximum depth limit for nested steps. src/agentic_fleet/app.py [306-314] -async def process_response(response: Any, collected_responses: List[str]) -> None: +async def process_response(response: Any, collected_responses: List[str], depth: int = 0, max_depth: int = 5) -> None: + if depth >= max_depth: + logger.warning("Maximum step nesting depth reached") + return try: async with cl.Step(name="Response Processing", type="process") as main_step: - # ... nested steps processing ... + # ... nested steps processing with depth + 1 ... Suggestion importance[1-10]: 8 Why: Adding a depth limit for nested steps is crucial to prevent potential infinite recursion and stack overflow errors in complex response processing scenarios.	8
	Handle empty input validation The stream_text function should handle potential empty text input to avoid splitting on empty string. Add input validation to handle this edge case. src/agentic_fleet/app.py [57-61] async def stream_text(text: str) -> AsyncGenerator[str, None]: + if not text: + return words = text.split() for i, word in enumerate(words): await asyncio.sleep(STREAM_DELAY) yield word + (" " if i < len(words) - 1 else "") Apply this suggestion Suggestion importance[1-10]: 7 Why: The suggestion adds important input validation to prevent potential errors when empty text is passed to the stream_text function. This improves robustness and error handling.	7

[qodo-merge-pro-for-open-source]

CI Feedback 🧐

(Feedback updated until commit 8366454)

A test triggered by this PR failed. Here is an AI-generated analysis of the failure:

Action: test

Failed stage: Run tests with coverage [❌]

Failed test name: tests/integration/test_app.py

Failure summary: The action failed because of missing Azure OpenAI credentials. The tests could not be executed because the required authentication credentials were not provided. Specifically: The code requires either an API key, Azure AD token, or token provider None of the required environment variables (AZURE_OPENAI_API_KEY or AZURE_OPENAI_AD_TOKEN) were set This affected multiple test files: test_app.py, test_cli.py, and test_models.py

Relevant error logs: 1: ##[group]Operating System 2: Ubuntu ... 1175: Python_ROOT_DIR: /opt/hostedtoolcache/Python/3.12.8/x64 1176: Python2_ROOT_DIR: /opt/hostedtoolcache/Python/3.12.8/x64 1177: Python3_ROOT_DIR: /opt/hostedtoolcache/Python/3.12.8/x64 1178: LD_LIBRARY_PATH: /opt/hostedtoolcache/Python/3.12.8/x64/lib 1179: ##[endgroup] 1180: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/pytest_asyncio/plugin.py:207: PytestDeprecationWarning: The configuration option "asyncio_default_fixture_loop_scope" is unset. 1181: The event loop scope for asynchronous fixtures will default to the fixture caching scope. Future versions of pytest-asyncio will default the loop scope for asynchronous fixtures to function scope. Set the default fixture loop scope explicitly in order to avoid unexpected behavior in the future. Valid fixture loop scopes are: "function", "class", "module", "package", "session" 1182: warnings.warn(PytestDeprecationWarning(_DEFAULT_FIXTURE_LOOP_SCOPE_UNSET)) 1183: ==================================== ERRORS ==================================== 1184: ________________ ERROR collecting tests/integration/test_app.py ________________ ... 1188: from .app import handle_message, initialize_session, update_settings # noqa: F401 1189: src/agentic_fleet/app.py:72: in <module> 1190: az_model_client = AzureOpenAIChatCompletionClient( 1191: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/autogen_ext/models/openai/_openai_client.py:1173: in __init__ 1192: client = _azure_openai_client_from_config(copied_args) 1193: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/autogen_ext/models/openai/_openai_client.py:102: in _azure_openai_client_from_config 1194: return AsyncAzureOpenAI(**azure_config) 1195: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/openai/lib/azure.py:435: in __init__ 1196: raise OpenAIError( 1197: E openai.OpenAIError: Missing credentials. Please pass one of `api_key`, `azure_ad_token`, `azure_ad_token_provider`, or the `AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_AD_TOKEN` environment variables. 1198: ___________________ ERROR collecting tests/unit/test_cli.py ____________________ ... 1202: from .app import handle_message, initialize_session, update_settings # noqa: F401 1203: src/agentic_fleet/app.py:72: in <module> 1204: az_model_client = AzureOpenAIChatCompletionClient( 1205: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/autogen_ext/models/openai/_openai_client.py:1173: in __init__ 1206: client = _azure_openai_client_from_config(copied_args) 1207: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/autogen_ext/models/openai/_openai_client.py:102: in _azure_openai_client_from_config 1208: return AsyncAzureOpenAI(**azure_config) 1209: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/openai/lib/azure.py:435: in __init__ 1210: raise OpenAIError( 1211: E openai.OpenAIError: Missing credentials. Please pass one of `api_key`, `azure_ad_token`, `azure_ad_token_provider`, or the `AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_AD_TOKEN` environment variables. 1212: __________________ ERROR collecting tests/unit/test_models.py __________________ ... 1216: from .app import handle_message, initialize_session, update_settings # noqa: F401 1217: src/agentic_fleet/app.py:72: in <module> 1218: az_model_client = AzureOpenAIChatCompletionClient( 1219: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/autogen_ext/models/openai/_openai_client.py:1173: in __init__ 1220: client = _azure_openai_client_from_config(copied_args) 1221: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/autogen_ext/models/openai/_openai_client.py:102: in _azure_openai_client_from_config 1222: return AsyncAzureOpenAI(**azure_config) 1223: /opt/hostedtoolcache/Python/3.12.8/x64/lib/python3.12/site-packages/openai/lib/azure.py:435: in __init__ 1224: raise OpenAIError( 1225: E openai.OpenAIError: Missing credentials. Please pass one of `api_key`, `azure_ad_token`, `azure_ad_token_provider`, or the `AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_AD_TOKEN` environment variables. ... 1238: src/agentic_fleet/cli.py 39 39 0 0 0% 3-69 1239: src/agentic_fleet/config.py 30 30 0 0 0% 3-58 1240: src/agentic_fleet/run.py 18 18 0 0 0% 3-31 1241: src/agentic_fleet/scripts.py 9 9 0 0 0% 2-14 1242: --------------------------------------------------------------------------- 1243: TOTAL 416 373 92 0 8% 1244: Coverage XML written to file coverage.xml 1245: =========================== short test summary info ============================ 1246: ERROR tests/integration/test_app.py - openai.OpenAIError: Missing credentials. Please pass one of `api_key`, `azure_ad_token`, `azure_ad_token_provider`, or the `AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_AD_TOKEN` environment variables. 1247: ERROR tests/unit/test_cli.py - openai.OpenAIError: Missing credentials. Please pass one of `api_key`, `azure_ad_token`, `azure_ad_token_provider`, or the `AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_AD_TOKEN` environment variables. 1248: ERROR tests/unit/test_models.py - openai.OpenAIError: Missing credentials. Please pass one of `api_key`, `azure_ad_token`, `azure_ad_token_provider`, or the `AZURE_OPENAI_API_KEY` or `AZURE_OPENAI_AD_TOKEN` environment variables. 1249: !!!!!!!!!!!!!!!!!!! Interrupted: 3 errors during collection !!!!!!!!!!!!!!!!!!!! 1250: 2 warnings, 3 errors in 8.92s 1251: ##[error]Process completed with exit code 2.