This example combines the simplest ASCII case, a precomposed accented BMP value and a supplementary emoji. It exercises every UTF-8 width except the three-byte range and proves that scalar, UTF-16 and byte counts answer different questions. The same tuple is the default calculator input and an automated acceptance case, so the article and tool cannot drift independently.

Load the mixed-width preset

Open the inspector and choose A · é · 😀, or paste the three values with no separators. Inspect mode preserves the exact textarea. The first row must be U+0041, decimal 65, UTF-8 41 and UTF-16 0041. The second must be U+00E9, decimal 233, UTF-8 C3 A9 and UTF-16 00E9. No locale rule changes either value.

Verify the supplementary row

The third row is one scalar, U+1F600, decimal 128512. Its UTF-8 sequence is F0 9F 98 80 and its UTF-16 sequence is D83D DE00. The escape column uses \u{1F600}, which names the scalar directly rather than presenting two independent surrogate escapes. The summary must show three scalars, four UTF-16 units and seven UTF-8 bytes.

Decode the copied U+ sequence

Copy U+0041 U+00E9 U+1F600, switch to code-points-to-text mode and run it. The reconstructed textarea must contain exactly Aé😀 and the evidence rows must match the original inspection. Adding commas or using 0x notation should fail instead of being repaired. This round trip proves the reviewed grammar and encoding, not that every external system will render the same glyphs.