What problem does it solve? Language models cannot access information beyond their training cutoff, making it impossible to answer questions about recent events, new publications, or current facts. This Skill connects to Perplexity models through OpenRouter to return real-time, web-grounded answers with source citations. ## Core Features & Use Cases - Real-time web search: Query five Perplexity models (Sonar, Sonar Pro, Sonar Pro Search, Sonar Reasoning, Sonar Reasoning Pro) through a single OpenRouter API key via LiteLLM. - CLI and programmatic access: Run searches from the command line with model, token, and temperature controls, or import search_with_perplexity as a Python module and save results to JSON. - Use Case: A researcher needs papers published in 2024 on CAR-T cell therapy for B-cell lymphoma. They run the search script with the sonar-pro model and receive a cited, up-to-date synthesis saved to a JSON file for downstream processing. ## Quick Start Set the OPENROUTER_API_KEY environment variable, install litellm with uv, then ask the assistant to search the web for a specific recent topic using the perplexity-search skill.
