Tool Gate · How it works
How the Model & System Card Builder gate works
How the Model & System Card Builder scores a card on completeness — then an evidence gate forces NOT PUBLISHABLE when the card makes a performance or fairness claim it doesn't back with named evidence, so an unevidenced claim overrides the score however complete the card looks. A card that makes no claim is graded on completeness alone.
How the gate works
This tool grades a model or system card you wrote, before you publish it. Each control is marked from your own evidence and weighted into a 0–100 score — and then the score is run past a gate that can only make the verdict worse.
The 7 weighted controls
| Signal | Weight |
|---|---|
| Intended use & users | 14 |
| Out-of-scope & prohibited uses | 16 |
| Training & evaluation data | 14 |
| Performance & metrics | 14 |
| Known limitations & failure modes | 16 |
| Fairness & bias considerations | 12 |
| Human oversight & contact | 14 |
| Total | 100 |
The verdict bands
A card that makes a performance or fairness claim it doesn't back with named evidence is NOT PUBLISHABLE however complete it looks. An unevidenced claim misleads the reader, so it overrides the completeness score — the gate can only make the verdict worse. Silence isn't dishonesty: a card that makes no claim is graded on completeness alone.
In the worked sample it scores 86 out of 100 and still reads NOT PUBLISHABLE — This card is 86% complete — six of seven sections are fully disclosed, and on the number alone it would read PUBLISHABLE. But it makes a performance claim while the performance section is only thin and the training-data provenance isn't fully disclosed, so the claim isn't backed by named evidence. An unevidenced claim overrides the score: the card is NOT PUBLISHABLE until the performance claim is substantiated or dropped. The one thing to fix first: Substantiate the performance claim with named evidence, or drop it — the card is complete enough to publish, but an unbacked claim keeps it NOT PUBLISHABLE..
Common questions
Can something score well on the How the Model & System Card gate works and still fail?
Yes. The weighted score is run past a worsen-only gate — the Unsubstantiated-claim — that can only make the verdict worse. In the worked sample it scores 86 out of 100 and still reads NOT PUBLISHABLE, because this card is 86% complete — six of seven sections are fully disclosed, and on the number alone it would read PUBLISHABLE. But it makes a performance claim while the performance section is only thin and the training-data provenance isn't fully disclosed, so the claim isn't backed by named evidence. An unevidenced claim overrides the score: the card is NOT PUBLISHABLE until the performance claim is substantiated or dropped.
What is a worsen-only gate?
A dispositive rule that can only lower a verdict, never raise it. One fatal flaw overrides an otherwise-good score, because a single disqualifying gap shouldn't hide behind a high average. The tool also names the one thing to fix first — here, "Substantiate the performance claim with named evidence, or drop it — the card is complete enough to publish, but an unbacked claim keeps it NOT PUBLISHABLE.".
A working aid, not legal advice. It grades the card you write from your own marks and claim declarations — it doesn't verify your disclosures are true, test the model, or guarantee any transparency regulation is met. It grades the card, not the model's quality and not any person. Confirm any mandatory transparency or instructions-for-use obligation with a qualified auditor or counsel.
Embed this diagram
Free to share and embed with attribution (CC BY 4.0) — keep the link to redhub.ai.
Interactive — renders the live diagram
<iframe src="https://redhub.ai/visuals/tool/model-system-card-builder.html" title="How the Model & System Card Builder gate works — RedHub AI" width="760" height="1844" loading="lazy" style="border:0;width:100%;max-width:760px"></iframe>
<p style="font:14px/1.5 system-ui,sans-serif"><a href="https://redhub.ai/visuals/tool/model-system-card-builder">How the Model & System Card Builder gate works</a> — by <a href="https://redhub.ai">RedHub AI</a>, the AI that tells you when to doubt it.</p>Image + link — a static picture for any blog
<a href="https://redhub.ai/visuals/tool/model-system-card-builder"><img src="https://redhub.ai/visuals/tool/model-system-card-builder-share.png" alt="Diagram of the Model & System Card Builder: a card scored to 0-100 and an evidence gate forcing NOT PUBLISHABLE when a claim isn't backed by named evidence." width="760" loading="lazy" style="max-width:100%;height:auto;border-radius:16px"></a>
<p style="font:14px/1.5 system-ui,sans-serif"><a href="https://redhub.ai">RedHub AI</a> — the AI that tells you when to doubt it.</p>
This is how Model & System Card Builder works. More diagrams in the Visual Field Guide.