router

class agentopera.router.AgentFactory[source]

Bases: object

classmethod create_agent_registry(intent_registry: IntentRegistry) → AgentRegistryBase[source]

Creates an AgentRegistry from an IntentRegistry.

Parameters:: intent_registry (IntentRegistry) – The shared config source.
Returns:: A runtime-compatible registry.
Return type:: AgentRegistryBase

classmethod create_intent_classifier(intent_registry: IntentRegistry) → LLMIntentClassifier[source]

Creates an LLM-powered intent classifier using the provided intent registry.

Parameters:: intent_registry (IntentRegistry) – The shared config source.
Returns:: Configured classifier using OpenAI-compatible tools.
Return type:: LLMIntentClassifier

classmethod create_intent_registry(intent_descriptions: dict[str, str], agent_intent_mapping: dict[str, str], agent_descriptions: dict[str, str], schema_overrides: Dict[str, dict] | None = None) → IntentRegistry[source]

class agentopera.router.AgentRegistry(intent_registry: IntentRegistry)[source]

Bases: AgentRegistryBase

Runtime-compatible agent registry adapter.

Wraps around IntentRegistry to expose methods required by SemanticRouterAgent and other runtime components.

Responsibilities: - Asynchronously fetch agents based on classified intent - Serve agent metadata for monitoring and validation

async get_agent(intent: str) → str[source]

Retrieves the agent type corresponding to a given intent.

Falls back to ‘chat’ if intent is not mapped.

Parameters:: intent – The intent for which to retrieve an agent.
Returns:: The agent type (e.g., ‘chat’, ‘vercel’, etc.)

async get_agents() → Set[str][source]

Retrieves the set of all registered agent IDs.

Returns:: A set of agent IDs.

async get_agent_description(agent_id: str) → str[source]

Retrieves the description of a specific agent by its ID.

Parameters:: agent_id – The agent ID for which to retrieve the description.
Returns:: The agent’s description.

async get_agent_descriptions() → Dict[str, str | list[str]][source]

Retrieves descriptions of all registered agents.

Returns:: A dictionary mapping agent IDs to their descriptions.

class agentopera.router.IntentRegistry(intent_descriptions: Dict[str, str], agent_intent_mapping: Dict[str, str], agent_descriptions: Dict[str, str | list[str]], schema_overrides: Dict[str, dict] | None = None)[source]

Bases: object

Internal configuration holder for intent and agent relationships.

This class is designed to manage: - Intent descriptions for classification models - Mapping from intent → agent_id - Mapping from agent_id → descriptions

It acts as the source of truth for the classifier and agent registry.

classmethod from_api_data(api_data: List[UserAgentInfo]) → IntentRegistry[source]

merge(other: IntentRegistry) → IntentRegistry[source]

get_all_intents() → Set[str][source]: Returns all available intent keys.

get_schema_override(intent: str) → dict[source]: Returns all overriding schemas if any

get_description(intent: str) → str[source]: Returns description for a specific intent.

get_agent_for_intent(intent: str) → str[source]: Returns agent_id associated with the given intent.

get_all_agents() → Set[str][source]: Returns all registered agent IDs.

get_agent_description(agent_id: str) → str | list[str][source]: Returns description for the specified agent.

get_tool_schemas() → list[dict][source]

Builds OpenAI-compatible tool definitions for all registered intents.

Returns:: List of function tool schema dictionaries for OpenAI-compatible models.

class agentopera.router.SemanticRouterAgent(name: str)[source]

Bases: RoutedAgent

async initialize()[source]

async route_to_agent(message: TextMessage | MultiModalMessage, ctx: MessageContext) → None[source]

event_log(session_id: str, agent_id: str) → None[source]

Logs the agent selection event to PostHog.

Parameters:

session_id (str) – The ID of the current customer session.
agent_id (str) – The ID of the agent handling the message.

async contact_agent(agent_name: str, message: TextMessage | MultiModalMessage, ctx: MessageContext, session_id: str) → None[source]

class agentopera.router.LLMIntentClassifier(intent_registry: IntentRegistry, model='Llama-3.1-8B-Instruct')[source]

Bases: IntentClassifierBase

async classify_intent(message: TextMessage | MultiModalMessage, timeout: float | None = 80.0) → str[source]: Returns the intent with the highest confidence score.