An awesome list that fact-checks the AI reverse-engineering hype

What it does An unofficial, bilingual awesome list tracking the wave of AI agents reverse-engineering and rebuilding closed software. Each case gets a tag: REA (the tool was used, verified), AI-RE (AI reverse engineering with other tools), Rebuild (clean-room rewrite, no RE evidence), or Claim? (unverified). The list indexes 37 runnable open-source projects by the app they replace, plus showcases, research, and legal notes.

The interesting bit It’s an awesome list with a skeptical spine. Of the six most-shared viral cases, only one — DX-Ball — says in its own repo that it used REA. The list also surfaces the two controlled studies of REA: 482 runs in rea-lab found no significant solve-rate gain (at roughly double the time and cost), and leanrea found a plain-CLI skill nearly matched REA’s accuracy for about 35% less cost. That’s rarer than it sounds — most awesome lists aggregate, this one audits.

Key highlights

• Every case tagged and checked, with a dedicated claim-check page (docs/is-it-rea.md)
• Index of 37 open-source replacements by the app they replace (OBS, Logitech G HUB, Plex on LG TVs, PowerToys)
• Includes the evidence on whether REA actually helps, not just the cheerleading
• Dates stamped on figures, since the ecosystem “moves weekly”
• CC0-licensed, with a legal section on RE, licenses, and terms of service

Caveats

• Numbers are snapshots (checked 2026-10-21, refreshed through 2026-11-05) and will age fast
• Unofficial and explicitly not affiliated with morluto/rea, despite the name overlap

Verdict Worth a star if you’re curious whether AI-driven reverse engineering is real or mostly viral screenshots — this list does the verification work others skip. If you just want RE tooling itself, go to the upstream repos it links.