research

name

research

description

Web research with Perplexity via OpenRouter. Use when you need to search the web and synthesize current information. Supports two modes: (1) Basic Sonar for general research and fact-checking, (2) Sonar Deep Research for complex analysis, multi-step queries, or when you're uncertain if deep research might be needed (ask for confirmation). Uses OpenRouter's API - no separate Perplexity key required.

OpenRouter Perplexity Research Skill

This skill enables web research using Perplexity models through OpenRouter, with two research depths available.

Models

• perplexity/sonar — Fast, general-purpose web research. Good for fact-checking, recent news, quick lookups.
• perplexity/sonar-deep-research — Deeper analysis with multi-step reasoning. Use for complex questions, comparative research, or investigation-style queries.

When to Suggest Deep Research

Before choosing a model, consider the query:

• Use basic Sonar for: "What's X's current stock price?", "Who won Y award?", "Latest news on Z"
• Suggest Deep Research for: Multi-part questions, "Compare X vs Y", "How would Z affect...", "What are implications of...", investigative queries

If unsure, suggest deep research and ask for confirmation. The user can always say "no, just use basic Sonar" and save tokens.

API Setup

The skill uses OpenRouter's API. Your OpenRouter API key must be configured in secrets.toml.

Making Requests

Call OpenRouter's completion endpoint with the Perplexity model ID:

POST https://openrouter.ai/api/v1/chat/completions

Headers:
  Authorization: Bearer {{secret:openrouter.api_key}}

Body:
{
  "model": "perplexity/sonar",  // or "perplexity/sonar-deep-research"
  "messages": [
    {
      "role": "user",
      "content": "Your research query here"
    }
  ]
}

Response structure:

• choices[0].message.content — Research result with citations
• usage.prompt_tokens, usage.completion_tokens — Token counts for tracking

Workflow

1. Parse the request — Is this a research task?
2. Pick the model — Basic Sonar for simple queries, suggest Deep Research for complex ones
3. Make the API call — Use http_request to hit the OpenRouter endpoint
4. Return results — Include citations and summary in your response

Cost Reference

• Sonar: $2/M input tokens, $8/M output tokens
• Sonar Deep Research: Same pricing, more reasoning

See references/pricing.md for full details.

Example: Basic Research

Query: "What's the current state of AI safety regulations in the EU?"

http_request(
  method: "POST",
  url: "https://openrouter.ai/api/v1/chat/completions",
  headers: {
    "Authorization": "Bearer {{secret:openrouter.api_key}}",
    "Content-Type": "application/json"
  },
  body: '{"model":"perplexity/sonar","messages":[{"role":"user","content":"What is the current state of AI safety regulations in the EU?"}]}'
)

Example: Complex Research (suggest deep research first)

Query: "Compare the business models and recent financial performance of OpenAI, Anthropic, and Mistral"

Suggestion to user: "This looks like a complex comparative analysis. Want me to use Deep Research for a more thorough investigation, or just quick research with basic Sonar?"

If confirmed, use perplexity/sonar-deep-research.