Web3 & Crypto News Intelligence Pipeline

This actor transforms RSS feeds from crypto news sources into structured, AI-analyzed intelligence. It scrapes articles, extracts content, and uses an LLM to generate summaries, sentiment, risk ratings, and categorization—all without requiring you to manage external API keys.

Overview

It functions as a combined crypto news scraper and analysis API. Instead of just fetching headlines, it processes full article content through an AI pipeline. The actor pulls from preset sources like CoinDesk and Cointelegraph, or a custom RSS feed. It first attempts direct scraping; if blocked, it automatically falls back to using the Brave Search API to retrieve content snippets. Each processed article results in a structured JSON object containing AI-derived insights.

Key Features

• No API Key Management: The infrastructure for the LLM (via OpenRouter) and fallback search (Brave) is bundled. You pay per successful analysis, not for subscriptions.
• AI-Powered Structuring: Every article is analyzed by an LLM to generate:
  • A concise executive summary.
  • A risk level (High, Medium, Low/Informational).
  • A category (e.g., DeFi Hack, Regulation, Protocol Upgrade).
  • Identified key entities (tokens, protocols, people).
• Resilient Content Extraction: Uses a "scrape-first" engine with automatic fallback to paid search to maximize content retrieval success.
• Structured JSON Output: Output is clean, formatted data ready for trading bots, dashboards, or other automated systems.
• Test Mode: Includes a setting to generate dummy data for pipeline testing without incurring costs.

How to Use

The actor is configured for a straightforward, plug-and-play run:

1. Select a Source: Choose a preset source (e.g., coindesk, ethereum-foundation, all) or provide a custom RSS feed URL.
2. Set Volume: Define the maxArticles to process from the feed to control run scope and cost.
3. Run: Start the actor. It will sequentially fetch, extract content, and analyze each article.
4. Optional: Use the Region/Time filter to scope fallback search results, or enable Run Test Mode for a dry run.

Input/Output

Input Configuration (via Apify console or API):

{
  "newsSource": "coindesk", // or "all", "custom", etc.
  "maxArticles": 10,
  "customRssUrl": "", // if source is "custom"
  "regionTimeFilter": "Past Week", // optional
  "testMode": false
}

Output Data Model:
The dataset contains items with the following core fields:

Pricing

The actor uses a Pay-per-event model:
* You are charged a small fee for each successful AI Article Analysis (an article-summary event).
* Standard Apify platform fees for compute time (actor run duration) also apply.
* No fee is charged for articles that fail analysis. This model covers the underlying LLM and search API costs we manage on your behalf.