A few weeks ago I got targeted by a fake recruiter scam. I wrote about it on LinkedIn and it reached far more people than I expected. Here is the full version, with what I learned and what I now do before running any take-home test.
How it started
The recruiter's LinkedIn profile was very convincing. We had mutual connections, the company was real, and he even sent me a presentation of the project I would be working on. It looked like a normal hiring process.
Then came the technical test: a repository to clone and run on my machine.
The detail I almost missed
Something felt off, so before cloning it I asked Claude to analyze the URL. It said the repository didn't exist. Even the AI didn't catch the trick.
I took a closer look and noticed that the repository name swapped a lowercase "l" for an uppercase "I". In the browser's default font, they look almost identical.
When I analyzed the real repository, I found the malware. I reported it to GitHub and they took it down within a minute. I also warned the friends who had the recruiter in their connections.
One week later: Binance
In early November, Binance published its incident report on the US$ 6.3 billion crypto heist. According to the report, as summarized by The Register, one of the entry points was an engineer who went through a fake hiring process. At the end of the interview, he was asked to download an "assignment" with a code repository, which ran on his work computer during setup. The FBI attributed the attack to North Korean hackers.
Same playbook that reached me: a convincing recruiter and a coding test to run.
What I do now before running a take-home test
- Don't trust the profile, even with mutual connections. Anyone can write any job title on LinkedIn. Check if the person shows up on the company's website and, if possible, confirm through another channel.
- Read the repository name slowly. Paste the URL into an editor with a monospaced font, where "l", "I" and "1" look different. Same goes for package names and domains.
- Check the account that owns the repo. An account created a few days ago, with no history and a single repository, is a red flag.
-
Read the code before running it. Especially anything that runs on its own: install scripts (
postinstallinpackage.json, for example),Makefile, git hooks and any command that downloads something from the internet. - Never run it on your work computer. If you must run it, use a disposable VM or container with no access to your credentials.
- Don't rely only on AI to review it. It helps, but in my case it missed the swapped letter too.
- If it's a scam, report it. On GitHub, use "Report repository" on the repo page. On LinkedIn, report the profile. It takes a minute and protects the next person.
Going further
I work at BIX Tecnologia, a Brazilian software company. We recorded a video (in Portuguese) with 10 ways to protect your project when coding with AI, Claude Code and GitHub:
Have you ever received a suspicious take-home test? Tell me in the comments how you spotted it. The more people know the playbook, the fewer fall for it.

Top comments (0)