agents.rag.self_reflective.agent¶

Self-Reflective Agentic RAG Agent.

from typing import Any Implementation of self-reflective RAG with critique and iterative improvement. Uses reflection loops to assess and enhance answer quality.

Classes¶

`ImprovedAnswer`	Result of answer improvement iteration.
`ReflectionCritique`	Structured critique from reflection.
`ReflectionPlan`	Plan for iterative improvement based on reflection.
`ReflectionType`	Types of reflection analysis.
`SelfReflectiveRAGAgent`	Self-Reflective RAG agent with iterative improvement capabilities.
`SelfReflectiveResult`	Complete result from self-reflective RAG process.

Functions¶

`create_self_reflective_rag_agent`(documents[, ...])	Create a Self-Reflective RAG agent.
`get_self_reflective_rag_io_schema`()	Get I/O schema for Self-Reflective RAG agents.

Module Contents¶

class agents.rag.self_reflective.agent.ImprovedAnswer(/, **data)¶

Bases: pydantic.BaseModel

Result of answer improvement iteration.

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.rag.self_reflective.agent.ReflectionCritique(/, **data)¶

Bases: pydantic.BaseModel

Structured critique from reflection.

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.rag.self_reflective.agent.ReflectionPlan(/, **data)¶

Bases: pydantic.BaseModel

Plan for iterative improvement based on reflection.

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.rag.self_reflective.agent.ReflectionType¶

Bases: str, enum.Enum

Types of reflection analysis.

Initialize self. See help(type(self)) for accurate signature.

class agents.rag.self_reflective.agent.SelfReflectiveRAGAgent¶

Bases: haive.agents.base.agent.Agent

Self-Reflective RAG agent with iterative improvement capabilities.

This agent uses conditional edges to iterate through reflection loops.

build_graph()¶

Build the self-reflective graph with conditional edges.

Return type:: haive.core.graph.state_graph.base_graph2.BaseGraph

classmethod from_documents(documents, llm_config=None, max_iterations=3, quality_threshold=0.85, **kwargs)¶

Create Self-Reflective RAG agent from documents.

Parameters:

documents (list[langchain_core.documents.Document]) – Documents to index for retrieval
llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
max_iterations (int) – Maximum reflection iterations
quality_threshold (float) – Quality threshold to stop iterating
**kwargs – Additional arguments

Returns:

SelfReflectiveRAGAgent instance

generate_initial_answer(state)¶

Generate initial answer.

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

improve_answer(state)¶

Improve the answer based on reflection.

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

reflect_and_critique(state)¶

Generate critiques and plan improvements.

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

setup_agent()¶

Initialize engines.

Return type:: None

should_continue_improving(state)¶

Determine if improvement should continue.

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

synthesize_result(state)¶

Create final self-reflective result.

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

class agents.rag.self_reflective.agent.SelfReflectiveResult(/, **data)¶

Bases: pydantic.BaseModel

Complete result from self-reflective RAG process.

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.rag.self_reflective.agent.create_self_reflective_rag_agent(documents, llm_config=None, reflection_mode='thorough', **kwargs)¶

Create a Self-Reflective RAG agent.

Parameters:

documents (list[langchain_core.documents.Document]) – Documents for retrieval
llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
reflection_mode (str) – Mode (“quick”, “balanced”, “thorough”)
**kwargs – Additional arguments

Returns:

Configured Self-Reflective RAG agent

Return type:

SelfReflectiveRAGAgent

agents.rag.self_reflective.agent.get_self_reflective_rag_io_schema()¶

Get I/O schema for Self-Reflective RAG agents.

Return type:: dict[str, list[str]]