Flake-setup.exe file from the releases section — no installer, no admin rights, no registration.On GitHub, someone published Tropicaldrescoop/discord-id-osint-checker — an app called Flake. The repo has 43 commits, ready-made builds, and a README with a detailed feature description. The author calls it free and local: no shared database, everything is computed on your machine.
The core is Discord's snowflake. The first 42 bits of any ID encode the moment of creation, so the date appears offline, with no network requests. Flake pulls the profile (nickname, avatar, banner, badges, color, server tag) through Discord's official API — using your own bot token, which you create yourself.
Then it gets interesting. You save the IDs of trusted people — moderators, "middlemen," friends — and every check is compared against that list: a copied display name, lookalike letters from other alphabets, a pixel-perfect avatar. For invites, the app remembers where a link led before, cross-checks the server's age, the inviter's age, and links in the description against the FishFish phishing list. There's a raid mode: you throw in a batch of IDs, Flake plots an account creation timeline and highlights spikes — entirely offline. Cases export to HTML or PDF, each snapshot has a SHA-256 hash, and cases are deleted after 30 days unless you save them.
The scam scheme is old: copy a well-known middleman's nickname, swap in a similar-looking letter, grab their avatar — and wait for someone to bite. Public checkers that answered the question "when was this account created" have mostly shut down, according to the author. Our assessment: the demand here is real, not made up — the "middleman" market in Discord runs on trust in a nickname, and it breaks with a single character.
What Flake changes: it moves the check from someone else's server to your laptop. You don't hand your contact list over to strangers and you don't hit the rate limits of a shared token. For a moderator of a large server, that's the difference between "trusted it at a glance" and "saw that the account was created 40 minutes ago and the invite description contains the same phishing link."
Who wins — moderators, traders with "middlemen," everyone sorting through raids. Who loses — the very clones and fly-by-night services that lived on inattention. But we wouldn't cast Flake as a superhero. The app shows the data; the decision to ban or not still belongs to a human.
Everything we know comes from the README and releases of a single repository. There's no public code audit, and the encryption of local data and "nothing leaves your PC" are, for now, the author's claims — verifiable only with a debugger. The Windows build is unsigned, so the last line of defense is SmartScreen. Separately: the freshness of the FishFish phishing list, the update frequency, and what happens to the project if the author disappears. If you're paranoid — build it from source yourself.
Prediction: within a quarter, the repository will get a portable build and a signed Windows binary — both are already on the author's roadmap, and without a signature, half the users won't make it to the "Run anyway" button.
Disclaimer: a single source — the README and releases of the repository itself; there has been no independent verification of the code or the claims.
Source: github.com · post in Telegram