Basic search integration

This example builds a complete search integration from scratch — environment setup, request handling, error recovery, and result parsing.

You need an API key to follow this guide. Get one from the API Console.

Setup

Store your API key as an environment variable rather than hardcoding it:

export ANDI_API_KEY="your-api-key"

# .env
ANDI_API_KEY=your-api-key

# .env
ANDI_API_KEY=your-api-key

Complete example

#!/bin/bash

# Search with error handling
response=$(curl -s -w "\n%{http_code}" \
  "https://api.andiai.com/api/v1/search?q=best+programming+languages&limit=5" \
  -H "x-api-key: $ANDI_API_KEY")

http_code=$(echo "$response" | tail -1)
body=$(echo "$response" | sed '$d')

if [ "$http_code" -eq 200 ]; then
  echo "$body" | jq '.results[] | {title, link, source}'
else
  echo "Error $http_code: $(echo "$body" | jq -r '.error')"
fi

import os
import requests

api_key = os.environ["ANDI_API_KEY"]

response = requests.get(
    "https://api.andiai.com/api/v1/search",
    params={"q": "best programming languages", "limit": 5},
    headers={"x-api-key": api_key},
)

if response.status_code == 200:
    data = response.json()
    for result in data["results"]:
        print(f"{result['title']}")
        print(f"  {result['link']}")
        print(f"  {result['desc'][:100]}...")
        print()
elif response.status_code == 429:
    retry_after = response.headers.get("Retry-After", 1)
    print(f"Rate limited. Retry after {retry_after}s")
else:
    error = response.json()
    print(f"Error {response.status_code}: {error['error']}")

const apiKey = process.env.ANDI_API_KEY;

const url = new URL("https://api.andiai.com/api/v1/search");
url.searchParams.set("q", "best programming languages");
url.searchParams.set("limit", "5");

const response = await fetch(url, {
  headers: { "x-api-key": apiKey },
});

if (response.ok) {
  const data = await response.json();
  for (const result of data.results) {
    console.log(result.title);
    console.log(`  ${result.link}`);
    console.log(`  ${result.desc.slice(0, 100)}...`);
    console.log();
  }
} else if (response.status === 429) {
  const retryAfter = response.headers.get("Retry-After") || 1;
  console.log(`Rate limited. Retry after ${retryAfter}s`);
} else {
  const error = await response.json();
  console.log(`Error ${response.status}: ${error.error}`);
}

How it works

API key from environment — loaded from ANDI_API_KEY, never hardcoded
Query parameters — q for the search query, limit to cap results at 5
Status code check — handle success, rate limiting, and other errors separately
Result parsing — each result has title, link, desc, and source

Variations

With text extracts

Add extracts=true to get longer text passages from each result page:

response = requests.get(
    "https://api.andiai.com/api/v1/search",
    params={"q": "best programming languages", "extracts": "true"},
    headers={"x-api-key": api_key},
)

data = response.json()
for result in data["results"]:
    if result.get("extracts"):
        print(f"{result['title']}: {result['extracts'][0][:200]}")

With domain filtering

Restrict results to specific sites:

response = requests.get(
    "https://api.andiai.com/api/v1/search",
    params={
        "q": "best programming languages",
        "includeDomains": "stackoverflow.com,github.com",
    },
    headers={"x-api-key": api_key},
)

Pinning deep mode

The default auto mode already escalates to deeper treatment automatically when a query needs it. Set searchMode=deep to guarantee that treatment on every call, accepting a 2–3 second response:

response = requests.get(
    "https://api.andiai.com/api/v1/search",
    params={
        "q": "best programming languages",
        "searchMode": "deep",
    },
    headers={"x-api-key": api_key},
)

Context format for LLMs

Get results as markdown text, ready to pass into an LLM prompt:

response = requests.get(
    "https://api.andiai.com/api/v1/search",
    params={"q": "best programming languages", "format": "context"},
    headers={"x-api-key": api_key},
)

# Response is markdown text, not JSON
context = response.text

Next steps

RAG pipeline

Use search results as context for an LLM.

Query parameters

Full parameter reference.

Error handling

Error codes and retry strategies.

Response format

Understand the response structure.

Getting started

Use cases

Setup

Complete example

How it works

Variations

With text extracts

With domain filtering

Pinning deep mode

Context format for LLMs

Next steps

RAG pipeline

Query parameters

Error handling

Response format

​Setup

​Complete example

​How it works

​Variations

​With text extracts

​With domain filtering

​Pinning deep mode

​Context format for LLMs

​Next steps

RAG pipeline

Query parameters

Error handling

Response format

Setup

Complete example

How it works

Variations

With text extracts

With domain filtering

Pinning deep mode

Context format for LLMs

Next steps