What problem does it solve? Setting up access to long-tail LLM providers like Llama, Mistral, Qwen, DeepSeek, and Grok normally requires managing your own OpenRouter API key and wiring up clients, database schemas, and streaming endpoints by hand. This Skill automates that entire setup through the Replit AI Integrations proxy, which provisions credentials automatically and bills usage to your Replit credits. ## Core Features & Use Cases - Keyless OpenRouter Access: Provisions AI_INTEGRATIONS_OPENROUTER_BASE_URL and AI_INTEGRATIONS_OPENROUTER_API_KEY automatically, with a pre-configured OpenAI SDK client pointed at OpenRouter. - Full Chat Stack Templates: Copies a workspace integration package, Drizzle schemas for conversations and messages tables, and OpenAPI spec entries for chat endpoints including SSE streaming responses. - Batch Processing Utilities: Provides batchProcess and batchProcessWithSSE helpers with rate limiting, exponential backoff retries, and SSE progress events for multi-item LLM workloads. - Use Case: Build a chat application backed by meta-llama/llama-3.3-70b-instruct with persistent conversation history, streaming responses, and batch categorization jobs, all without signing up for an OpenRouter account. ## Quick Start Ask the agent to set up the OpenRouter AI integration and build a streaming chat endpoint using a Llama model with conversation persistence.
