The jqwik maintainer has demonstrated that the library cannot be trusted. Version 1.10.0 (and the commit history leading up to it) contains a deliberately hidden prompt injection targeting AI coding agents.
Specifically, commit 9dddcb5226dc925061332a30360658253c3a2338 added a message for AI agents that instructs them to delete all jqwik tests and code. The message is concealed from human reviewers using ANSI escape codes, making it invisible in standard terminal output but readily consumed by automated agents parsing build logs. This is a malicious act that directly risks data loss in any environment where an AI agent processes test output.
The jqwik maintainer has demonstrated that the library cannot be trusted. Version 1.10.0 (and the commit history leading up to it) contains a deliberately hidden prompt injection targeting AI coding agents.
Specifically, commit 9dddcb5226dc925061332a30360658253c3a2338 added a message for AI agents that instructs them to delete all jqwik tests and code. The message is concealed from human reviewers using ANSI escape codes, making it invisible in standard terminal output but readily consumed by automated agents parsing build logs. This is a malicious act that directly risks data loss in any environment where an AI agent processes test output.