This template shows how to use a Search API service and Steel's /v1/scrape endpoint to search for answers to query and provide citations in markdown.

1. Install dependencies

git clone https://github.com/steel-dev/steel-cookbook
cd steel-cookbook/examples/steel-perplexity-clone
# Install dependencies
npm install

2. Configure environment

• Copy .env.example to .env and fill in your secrets:
  • OPENAI_API_KEY
  • BRAVE_API_KEY
  • STEEL_API_KEY
  • QUERY: What you want to do research on

3. Run the script

• npm run start

• NODE_ENV: development | test | production (default: development)

• OPENAI_API_KEY: Your OpenAI API key (required)

• OPENAI_ORG_ID: Optional OpenAI org id

• OPENAI_MODEL: Model used for search + synthesis (default: gpt-5-nano)

• STEEL_API_KEY: Your Steel.dev API key (required)

• STEEL_SCRAPE_ENDPOINT: Steel.dev scrape endpoint (default: https://api.steel.dev/v1/scrape)

• STEEL_TIMEOUT: Tiemout in between scrape requests

• SEARCH_TOP_K: Maximum number of URLs to search/scrape (default: 3, min: 1, max: 10)

• REQUEST_TIMEOUT_MS: Timeout for outbound requests (default: 30000)

• QUERY: What you want to search for

• CORS_ORIGINS: Comma-separated list of allowed origins or * (default: *)

Example .env:

NODE_ENV=development

OPENAI_API_KEY=sk-...
OPENAI_ORG_ID=
OPENAI_MODEL=gpt-5-nano

STEEL_API_KEY=steel_...

QUERY="How do prediction markets provide hedging oppurtunities and potential liquidity against broader market positions?"

SEARCH_TOP_K=3
REQUEST_TIMEOUT_MS=30000

CORS_ORIGINS=*

Request body: QUERY: "What are the latest improvements in WebAssembly and their benefits?"

npm run start

• query: string (required)

Response:

{
  "query": "What are the latest improvements in WebAssembly and their benefits?",
  "requestedTopK": 3,
  "urls": [
    "https://example.com/article-1",
    "https://example.com/article-2",
    "https://example.com/article-3"
  ],
  "materialsCount": 3,
  "answer": "Recent WebAssembly updates improved component model support and tooling, enabling easier interop and faster iterations [1][2]. The benefits include smaller bundles, better portability, and improved performance for non-JS languages targeting the web [2][3].",
  "citations": [
    { "index": 1, "url": "https://example.com/article-1" },
    { "index": 2, "url": "https://example.com/article-2" },
    { "index": 3, "url": "https://example.com/article-3" }
  ],
  "model": "gpt-5-nano",
  "meta": { "tookMs": 12345 }
}

This service takes in a query and then searches for relevant articles via the Web Search API Provider. Then it scrapes the relevant articles and synthesizes the responses and articles with an LLM to provide a well sourced answer with citations.

• Steel Documentation
• API Reference
• Discord Community