Firebase Genkit proporciona una forma estructurada de crear flujos de IA con herramientas. Agregar la búsqueda de Scavio como herramienta Genkit brinda a sus flujos acceso web en vivo para establecer bases, investigar y enriquecer los datos. Este tutorial define la herramienta, crea un flujo de búsqueda y la prueba de un extremo a otro en TypeScript.
Requisitos previos
- Nodo.js 18+
- Firebase Genkit instalado
- Una clave API de Scavio de scavio.dev
- Configuración del proyecto TypeScript
Guia paso a paso
Paso 1: Definir la herramienta de búsqueda Scavio para Genkit
Cree una definición de herramienta Genkit escrita para búsqueda web.
Python
import os, requests, json
# Python equivalent of the TypeScript Genkit tool
# (Genkit is TypeScript-first, but the API call is the same)
API_KEY = os.environ['SCAVIO_API_KEY']
SH = {'x-api-key': API_KEY, 'Content-Type': 'application/json'}
# Genkit tool schema (for reference when writing TypeScript)
genkit_tool_schema = {
'name': 'webSearch',
'description': 'Search the web for current information. Returns structured results.',
'inputSchema': {
'type': 'object',
'properties': {
'query': {'type': 'string', 'description': 'Search query'},
'platform': {'type': 'string', 'enum': ['google', 'reddit', 'youtube', 'amazon'],
'description': 'Platform to search'}
},
'required': ['query']
},
'outputSchema': {
'type': 'array',
'items': {
'type': 'object',
'properties': {
'title': {'type': 'string'},
'link': {'type': 'string'},
'snippet': {'type': 'string'}
}
}
}
}
print('Genkit tool schema (TypeScript reference):')
print(json.dumps(genkit_tool_schema, indent=2))Paso 2: Implementar la función de búsqueda
Cree la función de búsqueda real a la que llama la herramienta Genkit.
Python
def genkit_web_search(query, platform=None):
"""Implementation matching the Genkit tool schema."""
body = {'query': query, 'country_code': 'us'}
if platform and platform != 'google':
body['platform'] = platform
data = requests.post('https://api.scavio.dev/api/v1/search',
headers=SH, json=body).json()
return [{'title': r.get('title', ''), 'link': r.get('link', ''),
'snippet': r.get('snippet', '')} for r in data.get('organic_results', [])[:5]]
# TypeScript equivalent:
ts_code = """
// genkit-search-tool.ts
import { defineTool } from '@genkit-ai/ai';
import { z } from 'zod';
export const webSearch = defineTool(
{
name: 'webSearch',
description: 'Search the web for current information',
inputSchema: z.object({
query: z.string().describe('Search query'),
platform: z.enum(['google', 'reddit', 'youtube', 'amazon']).optional()
}),
outputSchema: z.array(z.object({
title: z.string(),
link: z.string(),
snippet: z.string()
}))
},
async ({ query, platform }) => {
const body = { query, country_code: 'us', ...(platform && { platform }) };
const resp = await fetch('https://api.scavio.dev/api/v1/search', {
method: 'POST',
headers: { 'x-api-key': process.env.SCAVIO_API_KEY, 'Content-Type': 'application/json' },
body: JSON.stringify(body)
});
const data = await resp.json();
return (data.organic_results || []).slice(0, 5).map(r => ({
title: r.title || '', link: r.link || '', snippet: r.snippet || ''
}));
}
);
"""
print(ts_code)
# Test the Python version
results = genkit_web_search('genkit firebase tutorial 2026')
print(f'Test: {len(results)} results')
for r in results[:2]:
print(f' {r["title"]}')Paso 3: Crear un flujo Genkit con búsqueda
Conecte la herramienta de búsqueda a un flujo de investigación completo de Genkit.
Python
def simulate_genkit_flow(topic):
"""Simulate a Genkit flow that uses search for research."""
print(f'\n=== Genkit Flow: Research "{topic}" ===')
# Step 1: Search for overview
overview = genkit_web_search(f'{topic} overview 2026')
print(f' [1] Overview search: {len(overview)} results')
# Step 2: Search for comparisons
comparisons = genkit_web_search(f'{topic} vs alternatives comparison')
print(f' [2] Comparison search: {len(comparisons)} results')
# Step 3: Search Reddit for opinions
opinions = genkit_web_search(f'{topic} review', platform='reddit')
print(f' [3] Reddit opinions: {len(opinions)} results')
# Compile results
all_sources = overview + comparisons + opinions
print(f'\n Total sources: {len(all_sources)}')
print(f' Cost: ${3 * 0.005:.3f} (3 search calls)')
print(f'\n Top findings:')
for s in all_sources[:5]:
print(f' - {s["title"][:55]}')
return all_sources
simulate_genkit_flow('Firebase Genkit')
simulate_genkit_flow('Vercel AI SDK')Ejemplo en Python
Python
import os, requests
SH = {'x-api-key': os.environ['SCAVIO_API_KEY'], 'Content-Type': 'application/json'}
def genkit_search(query, platform=None):
body = {'query': query, 'country_code': 'us'}
if platform: body['platform'] = platform
data = requests.post('https://api.scavio.dev/api/v1/search', headers=SH, json=body).json()
return data.get('organic_results', [])[:5]
results = genkit_search('genkit search tool 2026')
for r in results:
print(f'{r["title"]}: {r.get("link", "")}')
print('Cost: $0.005')Ejemplo en JavaScript
JavaScript
const SH = { 'x-api-key': process.env.SCAVIO_API_KEY, 'Content-Type': 'application/json' };
async function genkitSearch(query, platform) {
const body = { query, country_code: 'us', ...(platform && { platform }) };
const data = await fetch('https://api.scavio.dev/api/v1/search', {
method: 'POST', headers: SH, body: JSON.stringify(body)
}).then(r => r.json());
return (data.organic_results || []).slice(0, 5);
}
const results = await genkitSearch('genkit tutorial 2026');
results.forEach(r => console.log(r.title));Salida esperada
JSON
Genkit tool schema (TypeScript reference):
{
"name": "webSearch",
"description": "Search the web for current information..."
}
Test: 5 results
Getting Started with Firebase Genkit
Genkit AI Flows Tutorial 2026
=== Genkit Flow: Research "Firebase Genkit" ===
[1] Overview search: 5 results
[2] Comparison search: 5 results
[3] Reddit opinions: 5 results
Total sources: 15
Cost: $0.015 (3 search calls)