Three small API-backed Apify actors: Docker Hub image tracker, website lead extractor, SEC Form 4 scanner

# What these actors do

Docker Hub Image Tracker

This actor hits hub.docker.com/v2/repositories for a given repository and reports what actually changed for tags in a time window. Reported attributes include image digest, size, supported architectures, and last-pushed/last-pulled timestamps. Typical input looks like:

Developer note: the Docker Hub API uses an ordering parameter with reversed convention — ordering=last_updated sorts newest-first, and a - prefix yields oldest-first. The actor is useful when a tag string stays the same but the underlying image or architectures change.

Website Lead Extractor

Built with Crawlee's CheerioCrawler (no headless Chromium). You give it a starting domain and depth, and it crawls within that domain to collect emails, phone numbers, and social profile links (LinkedIn, X, Facebook, Instagram, GitHub) found in the raw HTML. Example input:

Tradeoff: because it only sees initial HTML, it won't extract content generated entirely client-side in SPAs without server-side rendering. For static and server-rendered marketing sites it's faster and avoids headless-browser flakiness and configuration.

Insider Trading Alert (SEC Form 4)

This actor reads EDGAR's filing feed and each filing's XML to surface Form 4 transactions for a ticker or market-wide stream of new filings. Inputs can filter by ticker, transaction type, and minimum transaction value. Example:

Practical bug discovered: some filers' software encode boolean-like flags differently across filings. For example, an officer/director flag may appear as '1' / '0' in some filings and as true / absent in others. A simple === '1' check will miss the alternate encodings, so parsing must accept multiple representations.

# Implementation choices and tradeoffs

• Use upstream APIs or feeds where possible. Each actor reads official endpoints (Docker Hub API, site HTML, EDGAR XML) rather than rendering and scraping full pages.

• Keep each actor narrow. Focusing on a single data source and a single outcome reduces edge cases and configuration complexity.

• Avoid headless browsers when they aren't necessary. The lead extractor's Cheerio approach is faster and more predictable for server-rendered sites, but it won't work for client-only content.

# Where each actor fits

• Use the Docker Hub actor to detect when a tag's underlying image changes (digest, size, architecture). It's helpful for monitoring image integrity or architecture support changes.

• Use the Insider Trading Alert to monitor recent Form 4 filings by ticker or to scan new filings market-wide for trades meeting value or type filters.

# Quick operational notes