google-news-api

Search Google News and return articles with titles, sources, snippets, dates, and links. Apify-backed n8n community node, pay-per-result, and usable as an AI agent tool.

Package Information

Released: 4 mo ago
Downloads: 11 weekly / 276 monthly
Latest Version: 0.1.1
Author: John Cole

Documentation

n8n-nodes-google-news-api

An n8n community node that searches Google News and returns structured articles: title, source, snippet, date, and link. It is backed by the Google News API on Apify and bills per result, so there are no subscriptions and no minimums.

Installation · Credentials · Operations · Output · Example workflows · Pricing · Resources

What it does

Give the node a query, and it returns one item per news article with the title, source, snippet, relative date, and link. It also works as an AI Agent tool, so an agent can pull the latest news on demand.

  • Search news by query, optionally scoped to a location
  • Localize with a country code and language code
  • Toggle safe search and fetch multiple pages
  • Choose how much data to return per article: Simplified, Raw, or Selected Fields

Installation

Follow the n8n community nodes installation guide:

  1. In n8n, open Settings > Community Nodes.
  2. Select Install.
  3. Enter n8n-nodes-google-news-api as the npm package name.
  4. Agree to the risks of using community nodes, then select Install.

After it installs, the Google News node appears in the nodes panel.

n8n Cloud only allows verified community nodes. Until this node is verified, install it on a self-hosted n8n instance.

Credentials

You need a free Apify account and an API token.

  1. Sign in to the Apify Console.
  2. Open Settings > Integrations and copy your Personal API token.
  3. In n8n, create a new Apify API credential and paste the token.
  4. Use the credential's Test button to confirm it works.

The node also supports Apify OAuth2 if you prefer to connect that way.

Operations

Article > Search returns news articles that match a query.

Parameter Description
Search Query The query to search news for. Required.
Location Location to run the search from. Optional.
Country Code / Language Code Localization, for example us and en.
Safe Search Active or Off.
Maximum Pages How many result pages to fetch.
Output How much data to return: Simplified, Raw, or Selected Fields.

Output

Each article is returned as its own n8n item. The Output parameter lets you choose how much to return:

  • Simplified (default): a compact object with position, title, source, snippet, date, and link. This mode is also used automatically when the node runs as an AI Agent tool, to keep responses small.
  • Raw: every field the API returns for each article, using the original field names below.
  • Selected Fields: pick exactly which fields to include.

Fields (Raw and Selected Fields)

Field Type Description
position integer Rank of the article on the page
title string Article headline
link string Article URL
source string Publisher name
snippet string Short excerpt
date string Relative publish time, for example 12 hours ago

Example workflows

1. Daily news digest by topic

  1. Schedule Trigger: run each morning.
  2. Google News: Search Query your topic, Output Simplified.
  3. Send Email or Slack: send the list of title, source, and link.

2. Brand or competitor monitoring

  1. Schedule Trigger: run hourly.
  2. Google News: Search Query a brand name.
  3. Remove Duplicates then Slack: alert on new coverage.

3. Let an AI Agent summarize the news

  1. AI Agent node.
  2. Attach Google News as a tool.
  3. Ask "What's the latest on electric vehicles?" The agent calls the node (in Simplified mode) and summarizes the headlines.

Pricing

This node calls the Google News API on Apify, which is billed pay-per-result: a small per-search fee (a few cents per page of results) plus a fraction of a cent per article returned, with no subscription and no minimums. Apify also includes a free monthly usage tier that covers typical volumes. See the Actor page for current rates.

Resources

License

MIT

Discussion