haive.agents.planning.llm_compiler.dag_models

Structured DAG models for LLM Compiler planning.

These Pydantic models are used as structured output from the planner LLM, replacing the text-based output parser. The LLM returns a proper DAGPlan directly, which is more reliable and leverages AugLLMConfig’s structured output capability.

Classes

DAGPlan	A directed acyclic graph of tasks to execute.
DAGTask	A single task in the execution DAG.
JoinerDecision	The joiner's decision after inspecting execution results.

Module Contents

class haive.agents.planning.llm_compiler.dag_models.DAGPlan(/, **data)

Bases: pydantic.BaseModel

A directed acyclic graph of tasks to execute.

The planner LLM outputs this structured plan. Tasks with no dependencies run in parallel. The last task should be ‘join’ to aggregate results.

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 haive.agents.planning.llm_compiler.dag_models.DAGTask(/, **data)

Bases: pydantic.BaseModel

A single task in the execution DAG.

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 haive.agents.planning.llm_compiler.dag_models.JoinerDecision(/, **data)

Bases: pydantic.BaseModel

The joiner’s decision after inspecting execution results.

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)