agents.rag.query_decomposition.agent¶

Query Decomposition Agents.

Modular agents for breaking down complex queries into manageable sub-queries. Can be plugged into any workflow with compatible I/O schemas.

Classes¶

`AdaptiveQueryDecomposerAgent`	Adaptive query decomposition that selects best strategy.
`ContextualDecomposition`	Context-aware query decomposition.
`ContextualQueryDecomposerAgent`	Context-aware query decomposition agent.
`HierarchicalDecomposition`	Hierarchical query decomposition with levels.
`HierarchicalQueryDecomposerAgent`	Hierarchical query decomposition agent.
`QueryDecomposerAgent`	Basic query decomposition agent.
`QueryDecomposition`	Complete query decomposition result.
`QueryType`	Types of queries for decomposition strategy.
`SubQuery`	Individual sub-query in decomposition.

Functions¶

`create_query_decomposer`([decomposer_type, llm_config])	Create a query decomposer agent.
`get_query_decomposer_io_schema`()	Get I/O schema for query decomposers for compatibility checking.

Module Contents¶

class agents.rag.query_decomposition.agent.AdaptiveQueryDecomposerAgent(llm_config=None, enable_fallback=True, **kwargs)¶

Bases: haive.agents.base.agent.Agent

Adaptive query decomposition that selects best strategy.

Initialize adaptive query decomposer.

Parameters:

llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
enable_fallback (bool) – Whether to fallback to simpler decomposition if needed
**kwargs – Additional agent arguments

build_graph()¶

Build adaptive decomposition graph.

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

class agents.rag.query_decomposition.agent.ContextualDecomposition(/, **data)¶

Bases: pydantic.BaseModel

Context-aware query decomposition.

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.query_decomposition.agent.ContextualQueryDecomposerAgent(llm_config=None, context_threshold=0.7, **kwargs)¶

Bases: haive.agents.base.agent.Agent

Context-aware query decomposition agent.

Initialize contextual query decomposer.

Parameters:

llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
context_threshold (float) – Threshold for context sufficiency
**kwargs – Additional agent arguments

build_graph()¶

Build contextual decomposition graph.

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

class agents.rag.query_decomposition.agent.HierarchicalDecomposition(/, **data)¶

Bases: pydantic.BaseModel

Hierarchical query decomposition with levels.

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.query_decomposition.agent.HierarchicalQueryDecomposerAgent(llm_config=None, max_levels=3, **kwargs)¶

Bases: haive.agents.base.agent.Agent

Hierarchical query decomposition agent.

Initialize hierarchical query decomposer.

Parameters:

llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
max_levels (int) – Maximum hierarchy levels
**kwargs – Additional agent arguments

build_graph()¶

Build hierarchical decomposition graph.

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

class agents.rag.query_decomposition.agent.QueryDecomposerAgent(llm_config=None, max_sub_queries=5, **kwargs)¶

Bases: haive.agents.base.agent.Agent

Basic query decomposition agent.

Initialize query decomposer.

Parameters:

llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
max_sub_queries (int) – Maximum number of sub-queries to generate
**kwargs – Additional agent arguments

build_graph()¶

Build query decomposition graph.

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

class agents.rag.query_decomposition.agent.QueryDecomposition(/, **data)¶

Bases: pydantic.BaseModel

Complete query decomposition result.

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.query_decomposition.agent.QueryType¶

Bases: str, enum.Enum

Types of queries for decomposition strategy.

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

class agents.rag.query_decomposition.agent.SubQuery(/, **data)¶

Bases: pydantic.BaseModel

Individual sub-query in decomposition.

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.query_decomposition.agent.create_query_decomposer(decomposer_type='basic', llm_config=None, **kwargs)¶

Create a query decomposer agent.

Parameters:

decomposer_type (Literal['basic', 'hierarchical', 'contextual', 'adaptive']) – Type of decomposer to create
llm_config (haive.core.models.llm.base.LLMConfig | None) – LLM configuration
**kwargs – Additional arguments

Returns:

Configured query decomposer agent

Return type:

haive.agents.base.agent.Agent

agents.rag.query_decomposition.agent.get_query_decomposer_io_schema()¶

Get I/O schema for query decomposers for compatibility checking.

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