This is related to https://github.com/rust-lang/rust-analyzer/issues/17011 and can probably be fixed by emitting `#[::core::prelude::v1::test]` from my macros. tokio test does this as well: https://github.com/tokio-rs/tokio/blob/a1acfd8c20475e8a878b012ac4aa90b361f004ae/tokio-macros/src/entry.rs#L365 ) Lets just hope that `#[::core::prelude::v1::test]` stays a no-op, when `harness=false` for that particular test.