Skill: llm-router

Version: 1.0.0

Intelligent LLM routing: select the best model for each request.

Routes to OpenAI, Anthropic, Ollama, or custom models based on:

- Task complexity

- Cost optimization

- Latency requirements

- Privacy constraints

Features:

- Automatic model selection based on task analysis

- Cost/quality tradeoff optimization

- Fallback chains for reliability

- Usage tracking and analytics

name: llm-router version: 1.0.0 description: Intelligent LLM routing with automatic model selection and fallbacks

credentials:

• name: OPENAI_API_KEY required: false description: OpenAI API key
• name: ANTHROPIC_API_KEY required: false description: Anthropic API key
• name: OPENROUTER_API_KEY required: false description: OpenRouter API key (access to 100+ models)

inputs:

• name: prompt type: string required: true description: The prompt to process
• name: task_type type: string default: auto enum: [auto, simple, complex, code, creative, analysis] description: Task type for routing
• name: priority type: string default: balanced enum: [cost, quality, speed, balanced] description: Optimization priority
• name: max_cost_cents type: integer default: 10 description: Maximum cost per request in cents
• name: require_local type: boolean default: false description: Require local/private model (no cloud)
• name: fallback_models type: array default: [] description: Fallback model chain

outputs:

• name: response type: string description: Model response
• name: model_used type: string description: Which model was selected
• name: routing_reason type: string description: Why this model was chosen
• name: cost type: object description: "{tokens, cost_cents}"
• name: metadata type: object description: Routing and execution metadata

backends:

• name: multi type: hybrid description: Routes to best available backend

components: inputs: [stdin, http_server] processors: [mapping, switch, openai_chat_completion, http, log] outputs: [stdout, sync_response]