agents.patterns.react_structured_reflection_patterns¶

Comprehensive ReactAgent → SimpleAgent Patterns with V3, V4, and Enhanced Base Agent.

This demonstrates all variations of ReactAgent → SimpleAgent workflows: 1. V3: ReactAgent → SimpleAgent (structured output) 2. V4: MultiAgent composition 3. Enhanced Base: Using enhanced base agent with hooks 4. Reflection: ReactAgent → SimpleAgentV3 → ReflectionAgent 5. Graded Reflection: ReactAgent → GradingAgent → SimpleAgentV3 → ReflectionAgent

Each pattern showcases the generalized hook system and different architectural approaches.

Classes¶

`ProblemAnalysis`	Structured problem analysis.
`ReactToStructuredV3`	V3 Pattern: ReactAgent → SimpleAgentV3 with structured output.
`ReactToStructuredV4`	V4 Pattern: MultiAgent with ReactAgent → SimpleAgentV3.
`ReactWithGradedReflection`	Graded Reflection Pattern: ReactAgent → GradingAgent → SimpleAgentV3 → ReflectionAgent.
`ReactWithReflection`	Enhanced Base Agent Pattern: ReactAgent → SimpleAgentV3 → ReflectionAgent.
`ResearchFindings`	Structured research findings.
`TaskAnalysis`	Analysis of a complex task.

Functions¶

`create_graded_reflection_pattern`([name, tools, ...])	Create graded reflection pattern.
`create_reflection_pattern`([name, tools, ...])	Create Enhanced Base Agent pattern with reflection.
`create_v3_pattern`([name, tools, structured_output_model])	Create V3 pattern: ReactAgent → SimpleAgentV3.
`create_v4_pattern`([name, tools, structured_output_model])	Create V4 pattern: MultiAgent composition.
`data_analysis`(data_description)	Analyze data and extract insights.
`example_graded_reflection_pattern`()	Example: Graded Reflection Pattern.
`example_reflection_pattern`()	Example: Enhanced Base Agent with Reflection.
`example_v3_pattern`()	Example: V3 Architecture Pattern.
`example_v4_pattern`()	Example: V4 Architecture Pattern.
`main`()	Run all pattern examples.
`stakeholder_analysis`(context)	Analyze stakeholders for a given context.
`web_research`(topic)	Research a topic using web sources.

Module Contents¶

class agents.patterns.react_structured_reflection_patterns.ProblemAnalysis(/, **data)¶

Bases: pydantic.BaseModel

Structured problem analysis.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

class agents.patterns.react_structured_reflection_patterns.ReactToStructuredV3(name, tools=None, structured_output_model=TaskAnalysis, reasoning_config=None, structuring_config=None)¶

V3 Pattern: ReactAgent → SimpleAgentV3 with structured output.

Init .

Parameters:

name (str) – [TODO: Add description]
tools (list | None) – [TODO: Add description]
structured_output_model (type[pydantic.BaseModel]) – [TODO: Add description]
reasoning_config (haive.core.engine.aug_llm.AugLLMConfig) – [TODO: Add description]
structuring_config (haive.core.engine.aug_llm.AugLLMConfig) – [TODO: Add description]

async arun(input_data)¶

Execute V3 pattern: ReactAgent → SimpleAgentV3.

Parameters:: input_data (str)
Return type:: pydantic.BaseModel

class agents.patterns.react_structured_reflection_patterns.ReactToStructuredV4(**data)¶

Bases: haive.agents.multi.agent.MultiAgent

V4 Pattern: MultiAgent with ReactAgent → SimpleAgentV3.

Init .

classmethod create_analysis_workflow(name='v4_analysis', tools=None, structured_output_model=TaskAnalysis)¶

Create V4 analysis workflow.

Parameters:

name (str)
tools (list | None)
structured_output_model (type[pydantic.BaseModel])

Return type:

ReactToStructuredV4

class agents.patterns.react_structured_reflection_patterns.ReactWithGradedReflection(name, tools=None, structured_output_model=TaskAnalysis)¶

Graded Reflection Pattern: ReactAgent → GradingAgent → SimpleAgentV3 → ReflectionAgent.

Init .

Parameters:

name (str) – [TODO: Add description]
tools (list | None) – [TODO: Add description]
structured_output_model (type[pydantic.BaseModel]) – [TODO: Add description]

async arun(input_data)¶

Execute Graded Reflection pattern.

Parameters:: input_data (str)
Return type:: dict[str, Any]

class agents.patterns.react_structured_reflection_patterns.ReactWithReflection(name, tools=None, structured_output_model=TaskAnalysis, reasoning_config=None, structuring_config=None)¶

Enhanced Base Agent Pattern: ReactAgent → SimpleAgentV3 → ReflectionAgent.

Init .

Parameters:

name (str) – [TODO: Add description]
tools (list | None) – [TODO: Add description]
structured_output_model (type[pydantic.BaseModel]) – [TODO: Add description]
reasoning_config (haive.core.engine.aug_llm.AugLLMConfig) – [TODO: Add description]
structuring_config (haive.core.engine.aug_llm.AugLLMConfig) – [TODO: Add description]

async arun(input_data)¶

Execute Enhanced Base Agent pattern with reflection.

Parameters:: input_data (str)
Return type:: dict[str, Any]

class agents.patterns.react_structured_reflection_patterns.ResearchFindings(/, **data)¶

Bases: pydantic.BaseModel

Structured research findings.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

class agents.patterns.react_structured_reflection_patterns.TaskAnalysis(/, **data)¶

Bases: pydantic.BaseModel

Analysis of a complex task.

Create a new model by parsing and validating input data from keyword arguments.

Raises [ValidationError][pydantic_core.ValidationError] if the input data cannot be validated to form a valid model.

self is explicitly positional-only to allow self as a field name.

Parameters:: data (Any)

agents.patterns.react_structured_reflection_patterns.create_graded_reflection_pattern(name='react_graded_reflection', tools=None, structured_output_model=TaskAnalysis)¶

Create graded reflection pattern.

Parameters:

name (str)
tools (list | None)
structured_output_model (type[pydantic.BaseModel])

Return type:

ReactWithGradedReflection

agents.patterns.react_structured_reflection_patterns.create_reflection_pattern(name='react_with_reflection', tools=None, structured_output_model=TaskAnalysis)¶

Create Enhanced Base Agent pattern with reflection.

Parameters:

name (str)
tools (list | None)
structured_output_model (type[pydantic.BaseModel])

Return type:

ReactWithReflection

agents.patterns.react_structured_reflection_patterns.create_v3_pattern(name='react_structured_v3', tools=None, structured_output_model=TaskAnalysis)¶

Create V3 pattern: ReactAgent → SimpleAgentV3.

Parameters:

name (str)
tools (list | None)
structured_output_model (type[pydantic.BaseModel])

Return type:

ReactToStructuredV3

agents.patterns.react_structured_reflection_patterns.create_v4_pattern(name='react_structured_v4', tools=None, structured_output_model=TaskAnalysis)¶

Create V4 pattern: MultiAgent composition.

Parameters:

name (str)
tools (list | None)
structured_output_model (type[pydantic.BaseModel])

Return type:

ReactToStructuredV4

agents.patterns.react_structured_reflection_patterns.data_analysis(data_description)¶

Analyze data and extract insights.

Parameters:: data_description (str)
Return type:: str

async agents.patterns.react_structured_reflection_patterns.example_graded_reflection_pattern()¶: Example: Graded Reflection Pattern.

async agents.patterns.react_structured_reflection_patterns.example_reflection_pattern()¶: Example: Enhanced Base Agent with Reflection.

async agents.patterns.react_structured_reflection_patterns.example_v3_pattern()¶: Example: V3 Architecture Pattern.

async agents.patterns.react_structured_reflection_patterns.example_v4_pattern()¶: Example: V4 Architecture Pattern.

async agents.patterns.react_structured_reflection_patterns.main()¶: Run all pattern examples.

agents.patterns.react_structured_reflection_patterns.stakeholder_analysis(context)¶

Analyze stakeholders for a given context.

Parameters:: context (str)
Return type:: str

agents.patterns.react_structured_reflection_patterns.web_research(topic)¶

Research a topic using web sources.

Parameters:: topic (str)
Return type:: str