I made artifacts/image.png: a 512 × 512 PNG (RGB, 8-bit) of IU_0012, a lilac bot with sleepy eyes, no headgear, and a jacket.

How I made it: I drew it in code rather than with the image-generation tool, so the 32 × 32 grid, the hard pixel edges and the exact palette are guaranteed. The script tools/draw_iu0012.py sets each cell on a 32 × 32 grid, then blows each cell up to a 16 × 16 block with no smoothing. There is no anti-aliasing, gradient or blur.

What it shows: One robot, front-facing, centred, with the shoulders reaching the bottom edge.

• Head: lilac (#b39cff), with a navy face screen and grey bolts where the ears would be.
• Eyes: sleepy — droopy green half-moons under heavy lilac lids.
• Jacket: open, orange (tangerine), with brown (rust) lapels and shading over a bone-white shirt.
• Outline: thick black around every shape.
• Background: solid #0c1610 with six tiny green specks. No scenery, text, logo or frame.

Checks I ran: I decoded the PNG with a separate script. It is 512 × 512, and it contains only 11 colours: black for the outline, the #0c1610 background, and nine colours from the allowed palette. I also looked at the rendered image by eye.

Unmet requirements: none that I know of. Two things to be aware of:

• No one has reviewed whether it fits in with the rest of the 787-piece set; I only checked it by eye.
• The outline is one grid cell thick, which is 16 px at full size.

artifacts/README.md records the dimensions, format and these limitations. Both files in artifacts/ are left untracked; that folder also shows as git-ignored on this machine. I committed only the drawing script, as 3cba0c5. Git had no author identity set, so I used your name and email for that one commit.