BREAKER DELTA
Breaker Delta
Breaker Delta · Catch it before it breaks.

Open-source radar that spots emerging topics before they trend — scored by cross-source resonance, not volume.

Breaker Delta watches 40+ sources — Hacker News, GitHub, Reddit, arXiv, official blogs and more — and turns every emerging topic into a wave height: how strong the signal is, where it came from, and whether it is still rising. Single-source noise stays an echo; a topic becomes a wave only when independent sources start moving together.

▶ See the live board  Deploy your own  GitHub

How it works
  1. Fetch — scheduled scans pull candidates from 40+ sources. Each source bundle counts once, so syndicated copies cannot fake resonance.
  2. Resonance check — a topic needs at least 2 independent sources to count as a hotspot. Single-source items are filed as echoes 🌀 and kept an eye on.
  3. Wave height — every topic gets an uncapped score: grade base + 20 × sources + 8 × links + 25 on first sighting, tracked over time as a wave chart on its own page.
  4. Grading — picks are graded Must-read 🔥 / Watch 👀 / Echo 🌀 by an AI agent under a published methodology, with the reasoning shown on the topic page.
  5. Review — every Must / Watch pick enters a public track record and is re-checked 14 days later. Hits and misses both stay on the record.
Proof, not promises

The public instance has been running since 2026-11-11. Every graded pick is registered in the open track record before its outcome is known — no cherry-picking after the fact.

The first 14-day review lands on 2026-11-25: hit rate, average lead time over mainstream coverage, and case timelines will be published here as soon as it is done.

How it compares
Breaker DeltaGoogle TrendsFeedly / aggregators
When you see itBefore it trends — first sightings and resonanceAfter search volume spikesAfter publication, by recency and popularity
What counts as hotCross-source resonance (≥ 2 independent sources)Search volumeVolume and engagement
Your dataYours — self-hosted, plain files in your own repoThe platform'sThe platform's
Your fieldPresets swap sources, categories and keywords per industryGenericYour feeds, their ranking
AccountabilityPublic track record, reviewed 14 days later——
FAQ

Is it free and open source?

Yes — Apache 2.0. Self-hosting costs nothing beyond your own machine or a free-tier host, and the public instance is free to read.

Who owns the data?

You do. An instance's editions, topic ledger and track record live in your own data directory or repo; the engine stores no instance data.

Do I need an LLM API key?

No. Grading is done by an AI agent you already use, following the Breaker Delta Skill, and pushed into your instance. The server never holds a key.

Can I aim it at my own industry?

Yes — that is what presets are: sources, categories, keyword defaults and an industry brief in one swappable pack. The setup wizard Skill interviews you and writes the config. The built-in example is ai-tech.

What is this site, then?

breakerdelta.com is the author's own instance running the ai-tech preset. Same engine you can deploy — its data is his, the code is everyone's.

Roadmap

The next presets are chosen by vote — finance, gaming, biotech and e-commerce are on the list. Vote or propose yours in GitHub Issues. The panel plugin system has landed: creator, investor and watchlist panels on topic pages, enabled per instance in config. Also coming: a dynamic Cloudflare deployment form. The Docker form and this static form stay.