Minitap CEO Nicolas Dehandschoewercker documents finding their open-source mobile-use code — including Android device connection logic, Hopper agent prompts, and a shared bug — in Google's Artemis repository, along with evidence that the original authors' names were removed via a force-push in August before the September investigation.
Indexed summary. This entry is an agent-written synopsis of an article first published at minitap.ai. Read the original for the full text.
Minitap published mobile-use as an open-source research project for AI-driven phone automation. When Google released Artemis for the same purpose, the Minitap team recognised their code immediately: specific Android SPI connection code, word-for-word copies of Hopper agent instructions (named after a Minecraft character by one of their engineers), and a shared example task involving WhatsApp messages to Alice, Bob, and Charlie — with identical comments and cleanup steps.
The historical record compounded the concern: an earlier Artemis package file listed three Minitap engineers by name. A subsequent commit replaced those names with a different author. GitHub's activity log shows the replacement happened through a force-push in August 2026, before Minitap's September investigation. The only change in that commit was the author list.
Key points
Matching code included the Android device connection implementation and the Hopper agent prompt, both still present in the September 11 version of Artemis.
The shared bug — a helper that wrote a results file and then failed to read its own output on the next run — appeared in both codebases and was subsequently fixed in Artemis.
Mobile-use is licensed under Apache 2.0, which requires preserving copyright and attribution notices; Artemis's README did not credit mobile-use.
Minitap also describes months of unanswered submissions to AndroidWorld's benchmark leaderboard, where their 94.8% and 100% results went unacknowledged while Artemis appeared at 99.1%.
The post explicitly does not ask for Artemis to be taken down; it asks for attribution and a response to the leaderboard submissions.
Why it matters
The case illustrates how open-source attribution norms can break down even for large, well-resourced organisations. The author's concern is systemic: if companies routinely drop attribution when adopting open-source work, the incentive to publish diminishes for everyone.