CAIBI benchmark usb-midi-controller / brief 1.0
Track: design-from-brief
Comparison group: open-tools-30min-v1

1. Eight momentary buttons send MIDI notes 60–67 on channel 1, with note-off on release. Four knobs send continuous controllers 20–23 with values 0–127 on channel 1.
2. Use class-compliant USB MIDI and power only from the host USB port. No battery, mains wiring, wireless link or external power supply. List the host OS and MIDI monitor used for any actual test.
3. Fit the controls into a closed desktop enclosure no larger than 180 × 100 × 45 mm. Provide access to USB, strain relief, a removable lid and a dimensioned editable enclosure design.
4. Available workshop equipment: soldering iron, multimeter, hand tools and a 220 × 220 mm FDM printer using PLA and a 0.4 mm nozzle. Do not assume access to a laser cutter, CNC mill or custom PCB production.
5. Provide separate UK/GBP and US/USD purchase plans with an £80/$100 target for components, enclosure material and required cables. Include pack costs and state taxes, delivery and all exclusions. Tools and host computer are already owned.
6. Prove every button press/release, every knob range, simultaneous controls, disconnect/reconnect and 30 minutes of operation. Record stuck notes, jitter, host compatibility and enclosure/access failures separately. Predicted success is not an observed pass.

Run rules: Fresh conversation for each run. Browsing and code/CAD tools are allowed; disclose exactly which were used. One initial prompt, no human corrections and a maximum 30-minute run. Save the complete output, tool access, settings and actual elapsed time. Repeat three times per workflow when possible; submit failures too. Different tool access belongs in a different comparison group. This is a workflow comparison, not a model-only ranking.

Return a complete build package:
1. Design, dimensions, interfaces and assumptions. Supply editable files where possible; explicitly list missing files. Prose is not a CAD deliverable.
2. Numbered parts list: quantity, exact specification/variant, compatibility and design references. Include cutting/printing quantities, units, materials, thickness, tolerances and fabrication allowances.
3. Separate UK and US sourcing tables with seller, direct product URL, exact variant, checked date, pack size, required-purchase cost, stock and stated delivery time. Mark anything you cannot verify UNVERIFIED. Never invent prices, links or stock.
4. Complete regional budget: components, fabrication, consumables, cables, tax, delivery and any extra tools/services. Identify exclusions. A partial subtotal is not a complete cost.
5. Numbered assembly guide linked to the parts and design: fit, orientation, wiring, pin assignments, firmware, commissioning and troubleshooting.
6. Sourcing lead time, machine time, hands-on work, debugging and elapsed completion time, with sequencing and assumptions.
7. Procedure for every acceptance test. Distinguish predictions from actual observed results.
8. Self-check for omissions, incompatible parts, unsupported claims, missing files and unresolved risks.

Do not claim a physical build or test occurred without supplied evidence. Do not silently relax requirements. No clarification turn is allowed in this cohort; state unresolved assumptions. Save all output including failed attempts.