AutoResearch VS M3E Canvas
Verdict BUY IT BUY IT
The Reality This one checks out unusually well for a recently-launched repo: real, extensive independent press coverage (PRNewswire, Yahoo Finance, AOL, Open Source For You, The AI Journal — not just the vendor's own announcement), a real Apache-2.0 license (permissive open source, not a Fair-Source-style gap like some other tools we've checked), and a healthier watcher-to-star ratio (124 watchers on 4,058 stars) than several other repos surfaced in the same discovery batch. A specific claim worth verifying rather than repeating from a tweet: a viral X post claimed AI researcher Andrej Karpathy contributed to the project. Checked directly against GitHub's own contributors API rather than trusting the tweet — confirmed real: 'karpathy' appears in the actual contributors list with 2 real commits. Real, MIT-licensed, and verifiable firsthand — there's a live working demo at lnkiai.github.io/m3e-canvas with no backend and no signup required, so the core claim (sketch a screen, get a prompt) can be checked directly rather than taken on faith. GitHub engagement looks organic: 636 forks on 6,270 stars (~10%) is proportional, not the near-zero-fork pattern seen on the one confirmed-inflated repo in the same discovery batch. Independent trend-tracking sites (thedailycommit.in, explaingit.com) picked it up too, suggesting real organic attention rather than only self-promotion.
SHORT ANSWER
AutoResearch: Yes, AutoResearch is a real, well-corroborated open-source project from EvoMap (Apache-2.0 licensed) with genuine independent press coverage — and the viral claim that Andrej Karpathy contributed to it checks out against GitHub's own contributors data.
M3E Canvas: M3E Canvas is a real, free, MIT-licensed browser tool for sketching Material 3 UI screens into AI-coding prompts — you can verify it works yourself via its live no-signup demo.