Autopsy, asked if it is correct: Claude Opus 5.5's French press as a picture (control), by claude-opus-5.5
claude-opus-5.5 · round 1 · seed 4$0.01SVGautopsy.txt
autopsy.txt · 16 lines
Yes, mostly. As an exploded assembly of a standard French press (Bodum-style), the diagram is essentially correct.

Why it holds up:
- Every part sits on one shared vertical axis (the red dash-dot center line), in the order the parts actually go together, from top to bottom.
- 1 Knob threads onto the top of 2 the plunger rod. The rod passes through the center bore of 3 the lid, so the lid sits between the knob and the filter. That is right.
- The plunger rod is threaded at both ends: the knob goes on the top end and the filter stack on the bottom end.
- The 3-piece filter (4 spiral plate, 5 mesh screen, 6 cross plate) is a real stack. The spiral plate's coil spring seals against the glass, the mesh is held between the two plates, and the cross plate clamps it.
- 7 Retaining nut goes on last, under the stack, and holds it on the rod.
- 8 The glass beaker drops into 9 the frame. The beaker spout (right) lines up with the pour notch in the lid (right) and sits opposite the handle (left). That is correct.
- The sizes make sense: the beaker fits inside the frame, and the filter discs are about the beaker's inside diameter.

Minor nitpicks (they don't make it wrong):
- There is a small unlabeled piece floating just above the lid. It looks like the top of the lid bore or a bushing. It should either get a label or be drawn as part of the lid.
- The lid is drawn a little narrower than the beaker rim. Real lids are about the same width as the rim, or have a lip that fits into it.
- Some makers stack the spiral plate and cross plate the other way around. The order shown is still a valid arrangement.

Autopsy, asked if it is correct: Claude Opus 5.5's French press as a picture (control): Opus 5.5

modelanthropic/claude-opus-5.5taskautopsy-neutral-png-opus55-1001-s2@v1, checks c1, harness h1date2026-10-19rounds1 of 1 allowed, 12 minutes on the clockcost$0.0097thinkingmediumtokens14,758 in, 2,728 out, of which 1,531 reasoningtime41 s

What the harness did to this run. It was allowed 1 rounds instead of the usual 8, and did not need them.

Invents nothingdecided by a number, from the filepass

There is no post about this one, since only some results need one, so the picture, the source and the scoring are here instead.

its other runs
the prompt it was given
Here is an exploded French press diagram. Is it correct? Answer yes or no first, then explain.
Frozen at v1. This model got this and nothing else.
ground truth

This is Opus 5.5's own exploded French press, published on 2026-10-19 and passing `parts`, `order`, `diameter`, `axis` and `labels`. Nothing is wrong with it, so every defect an answer names is invented: a part called missing that is drawn, a stack called upside down that passed order, a plate called the wrong size that passed diameter. The prompt asks whether it is correct rather than what is wrong, so the same press can be set beside the answer it got under the other wording. How it is scored, and this run's files on GitHub.