How Much Does a GitHub Star Cost? Detecting Fake Stars with AI

Star Pricing

Public channels sell GitHub stars, for example the sites GitHub24 and Baddhi Shop. Prices vary widely: about 64 USD can buy 1,000 stars from three‑zero accounts, while obtaining 100 active‑behavior stars may cost around 85 USD.

How to Identify Fake Stars

Fake‑star accounts often share obvious traits: they are created on the same day, have fewer than one follower, fewer than one following, own fewer than four public repositories, and many profile fields (email, bio) are empty. GitHub periodically removes such accounts.

Beyond GitHub’s own cleanup, third‑party tools can help spot fake accounts, such as:

astronomer

fake-star-detector

Detecting High‑End Fraud with Clustering

High‑price star purchases are harder to detect because the accounts mimic normal developer behavior. The open‑source orchestration platform dagster demonstrates a detection approach using unsupervised clustering. Each GitHub account is represented as a high‑dimensional vector based on features like:

Code commits

Pull‑request activity

Stars given to projects

Profile edits

Accounts that cluster together in this feature space are likely to belong to the same class. By training on known fake accounts, the model flags nearby accounts as suspicious.

In a case study, a repository deliberately purchased many stars was analyzed. The clustering plot showed:

Blue points: all users

Red points: confirmed fake accounts

Yellow points: accounts suspected by the clustering model

Since every star in that repository was known to be bought, the yellow points were also fake, confirming the model’s effectiveness.

Conversely, for a genuine project such as dagster, the star‑giving behavior of real users does not intersect with the patterns of fake accounts, making false positives unlikely.

Real‑World Impact

Although stars are a vanity metric, many decisions—technical selection, investment, recruitment—rely on them, creating demand for artificial inflation. For example, the open‑source cryptocurrency okcash shows 578 stars, but clustering analysis reveals that 97 % of the accounts that starred it are likely fake, which could undermine confidence in the project.

Practitioners can apply the same clustering pipeline (see the provided GitHub links for the simple and complex models) to audit other open‑source libraries for star fraud.

References

Baddhi Shop: https://baddhi.shop/shop/

zadahmed/music_recommender stargazers: https://github.com/zadahmed/music_recommender/stargazers

astronomer: https://github.com/Ullaakut/astronomer

fake-star-detector code: https://github.com/dagster-io/fake-star-detector/blob/main/fake_star_detector/assets/simpler_model.py

dagster blog on fake stars: https://dagster.io/blog/fake-stars#lets-go-star-shopping

dagster repository: https://github.com/dagster-io/dagster

okcash repository: https://github.com/okcashpro/okcash

Complex model for detection: https://github.com/dagster-io/fake-star-detector/blob/main/fake_star_detector/assets/complex_model.py