Skip to content
← Leaderboard
THE RULES OF ENGAGEMENT

A fair score.
A very unfair roast.

Astra’s scorecard covers five parts of a board. Each carries the same weight, and the board’s score out of 100 is their average, rounded to the nearest five. Astra first identifies the board’s likely category and complexity, then applies the engineering standards that matter for that kind of design. Complexity sets context, not points.

01

Routing

Signal and interconnect geometry: topology, widths, clearances, vias, directness, congestion, unrouted copper, and return-path continuity.

20%
02

Power integrity

Power architecture and delivery: regulation, protection, decoupling, grounding, current capacity, rail distribution, and power-related thermal risk.

20%
03

Circuit design & placement

Non-power circuit design and component arrangement: logical connectivity, interfaces, topology, part use, functional grouping, proximity, orientation, mechanical alignment, and assembly access.

20%
04

Manufacturability

Fabrication, assembly, and testability: board-edge, drill, mask, courtyard, pad and footprint constraints, process fit, and production access.

20%
05

Clarity & silkscreen

On-board and project clarity: readable references, polarity and connector markings, naming, schematic organization, and useful build documentation.

20%
85–100 · Excellent or exceptional60–84 · Mixed to solid0–59 · Serious weaknesses

What a score means

95–100

Exceptional execution with no meaningful weakness. Rare.

85–94

Excellent execution with only minor issues.

70–84

Solid execution with notable improvements still needed.

50–69

Mixed execution with important weaknesses.

25–49

Poor execution with serious defects or confused choices.

0–24

Fundamentally broken or unsafe in this area.

Every finding and every area mark cites its evidence, and a claim Astra cannot cite is dropped rather than softened. An area Astra cannot assess contributes a neutral 50, so missing evidence cannot raise the score by shrinking the average or lower it as a fake failure. Unresolved DRC and ERC errors do not cap the score — the counts ride beside it instead, so a board is never marked clean quietly. Humour, ambition, complexity and GitHub stars earn nothing. Identical files reuse their result, so a score cannot be rerolled.
I can take it. Submit a board