@radzor/agent-router

Route prompts to different AI models or agents based on intent classification. Supports static rule-based routing, LLM-powered intent detection, and fallback chains for reliable multi-agent orchestration.

AI & MLv0.1.0typescriptpythonServeragentrouterintentmulti-modelorchestrationaillmroutingby Radzor

Install

View source on GitHub →

$npx radzor@latest add agent-router

⚠ Constraints: Requires an OpenAI-compatible API key for intent classification. Agent handlers are user-provided async functions. Uses native fetch (Node.js 18+).

Inputs

Name	Type	Default	Description
apiKey*	string	—	API key for the classifier LLM used to detect intent (OpenAI-compatible).OPENAI_API_KEY
classifierModel	string	gpt-4o-mini	Model to use for intent classification.
baseUrl	string	https://api.openai.com/v1	Base URL for the classifier API.
fallbackAgent	string	—	Name of the agent to use when no intent matches. If not set, routing throws on no match.

apiKey*string

API key for the classifier LLM used to detect intent (OpenAI-compatible).

OPENAI_API_KEY

{ "llm": { "constraints": "Requires an OpenAI-compatible API key for intent classification. Agent handlers are user-provided async functions. Uses native fetch (Node.js 18+).", "usageExamples": "llm/examples.md", "integrationPrompt": "llm/integration.md" }, "name": "@radzor/agent-router", "tags": [ "agent", "router", "intent", "multi-model", "orchestration", "ai", "llm", "routing" ], "events": [ { "name": "onRouted", "payload": { "agent": "string", "intent": "string", "prompt": "string", "confidence": "number" }, "description": "Fired when a prompt is successfully routed to an agent." }, { "name": "onFallback", "payload": { "agent": "string", "prompt": "string", "reason": "string" }, "description": "Fired when routing falls back to the default agent." } ], "inputs": [ { "name": "apiKey", "type": "string", "envVar": "OPENAI_API_KEY", "required": true, "description": "API key for the classifier LLM used to detect intent (OpenAI-compatible)." }, { "name": "classifierModel", "type": "string", "default": "gpt-4o-mini", "required": false, "description": "Model to use for intent classification." }, { "name": "baseUrl", "type": "string", "default": "https://api.openai.com/v1", "required": false, "description": "Base URL for the classifier API." }, { "name": "fallbackAgent", "type": "string", "required": false, "description": "Name of the agent to use when no intent matches. If not set, routing throws on no match." } ], "radzor": "1.0.0", "actions": [ { "name": "route", "params": [ { "name": "prompt", "type": "string", "description": "The user prompt to route." } ], "returns": "Promise<RouteResult>", "description": "Classify the prompt's intent and route it to the best matching agent." }, { "name": "registerAgent", "params": [ { "name": "name", "type": "string", "description": "Unique name for this agent." }, { "name": "handler", "type": "(prompt: string) => Promise<string>", "description": "Async function that takes a prompt and returns a response." }, { "name": "intents", "type": "string[]", "description": "List of intent labels this agent handles." }, { "name": "description", "type": "string", "required": false, "description": "Description of what this agent does (used by the LLM classifier)." } ], "returns": "void", "description": "Register an agent with its intents and handler function." }, { "name": "removeAgent", "params": [ { "name": "name", "type": "string", "description": "Name of the agent to remove." } ], "returns": "boolean", "description": "Remove a registered agent. Returns true if the agent existed." }, { "name": "listAgents", "params": [], "returns": "AgentInfo[]", "description": "List all registered agents with their intents and descriptions." } ], "outputs": [ { "name": "routeResult", "type": "RouteResult", "fields": [ { "name": "selectedAgent", "type": "string", "description": "Name of the agent that handled the prompt." }, { "name": "confidence", "type": "number", "description": "Confidence score of the intent classification (0-1)." }, { "name": "response", "type": "string", "description": "The response from the selected agent." }, { "name": "intent", "type": "string", "description": "The detected intent label." } ], "description": "The result of routing a prompt to an agent." } ], "runtime": "server", "version": "0.1.0", "category": "ai", "languages": [ "typescript", "python" ], "description": "Route prompts to different AI models or agents based on intent classification. Supports static rule-based routing, LLM-powered intent detection, and fallback chains for reliable multi-agent orchestration.", "dependencies": { "radzor": [], "packages": {} }, "composability": { "connectsTo": [ { "output": "routeResult", "mapField": "response", "compatibleWith": [ "@radzor/llm-completion.action.complete.prompt", "@radzor/structured-output.action.generate.prompt" ] }, { "event": "onFallback", "description": "Alert when routing falls back to default agent", "compatibleWith": [ "@radzor/error-tracker.action.captureMessage.message", "@radzor/log-aggregator.action.warn.message" ] } ] } }

@radzor/agent-router

Install

Inputs

Outputs

Actions

Events

Composability

radzor.manifest.json