Open-Source Automation for Low-Volume Directory Scraping Scaling to High Load

Research Question

The user seeks a fully open-source, forkable platform for weekly scraping from 3 sources, deduplication, GPT-4o-mini description generation, Postgres storage, and tweeting new listings on a side project earning $0-500/month initially, scaling to 10k executions/day on a $20 VPS, comparing Activepieces, n8n, Windmill, and custom Python on Raspberry Pi with cron.

Key Findings

• All platforms (Activepieces, n8n, Windmill) are open-source, self-hostable at $0/month for 1 workflow, 7 steps, 4 executions/month initially and 300k executions/month at scale; fit free tiers (n8n Community, Windmill Free and Open-source, Activepieces unlimited self-hosted).
• Native integrations: n8n (OpenAI: 15 actions like "Message a model"; Postgres: 6 actions +1 trigger like "Insert rows"; Twitter/X: 7 actions like "Create Tweet"); Activepieces (OpenAI: 9 actions like "Ask ChatGPT"; Postgres: 1 action +1 trigger; Twitter: 2 actions like "Create Tweet"); Windmill (OpenAI: 58 actions; Postgres: 4 actions like "Execute Query"; no native Twitter—use HTTP).
• Community activity: n8n largest (51k GitHub stars, 9,297 templates; forum.n8n.io and GitHub issues/discussions with maintainer responses); Windmill (12k stars; Discord community support); Activepieces (10k stars; GitHub inferred).
• n8n's fair-code/Sustainable Use License allows forking/modifying for solo indie use; no restrictions on personal low-revenue projects.
• Scaling on single $20 VPS (~2vCPU/4GB): n8n breaks first (lacks free-tier queue_mode/multi-main/worker view, Node.js concurrency limits); Activepieces/Windmill viable longer (built-in workers, unlimited free self-hosted).
• Custom Python+cron: Fully open-source, no UI, requires full coding (BeautifulSoup/Scrapy, psycopg2, openai lib, Tweepy); lowest overhead but highest maintenance.

Recommendation

Use self-hosted n8n Community edition for its mature visual builder, richest integrations/community for initial low-volume setup, and confirmed license compatibility. Switch to Windmill if scaling to 10k executions/day soon, due to better free-tier concurrency/workers on a $20 VPS. Caveats: Test scraping (use HTTP/Code nodes); assumes 7 steps/workflow, single-user, self-hosted; add Redis queueing for n8n at scale.

Details

Pricing pages: n8n https://n8n.io/pricing ($0 Community self-hosted); Activepieces https://www.activepieces.com/pricing (pricing data unavailable, $0 self-hosted); Windmill https://www.windmill.dev/pricing ($0 Free and Open-source self-hosted). Prices as of 2026-04-20.