qalarc.com / projects / 4chan-scrape-data-analysis
chanalyse — 4chan Intelligence Engine
Scrapes 4chan boards 24/7, LLM-classifies threads into themes/stories, detects spikes, auto-writes sourced briefings, and publishes market-sentiment signals
chanalyse is a full discussion-intelligence pipeline: it continuously scrapes public 4chan boards (/biz/, /g/, /pol/), captures every thread and post permanently, and uses a 5-backend LLM router (Claude OAuth → Codex CLI → Claude CLI → API → heuristic) to classify each thread into an emergent theme → story → entity model. Z-score spike detection fires 'article triggers' that the engine researches (gathering real quotes and claims from the threads) and turns into neutral, provenance-tracked briefings — every claim traceable to its source posts, human-reviewed before publishing. A GDELT-style signals layer aggregates tone (−100…+100), mention-volume z-scores, momentum, dispersion and a composite buzz index per company/ticker/crypto — exposed as JSON/CSV anyone can consume. Live snapshot, briefings and callable signal data at qalarc.com/chanalyse.
Now live on its own domain: chanalyse.org — the site moved off qalarc.com in August 2026 to its own Cloudflare Pages project.
Capabilities
24/7 adaptive low-impact scraping with permanent raw archive
Emergent theme → story → entity LLM classification (5-backend router)
Z-score spike detection that fires article triggers
Automatic briefing writer with provenance-tracked research notes
GDELT-style market signals: tone, buzz index, momentum, confidence
Callable signal data — JSON/CSV at qalarc.com/chanalyse/data/
Entity trend index (Google-Trends-style 0–100) + movers detection
Static $0/month publishable site + private ops PWA + MCP server
Tags
Status: live · First built: 2024-03-01 · Last updated: 2026-08-25