Trust Score Badges
Add a trust score badge to your model's README
Badge Styles
Show your model's security posture with a shields.io-style badge. Available in multiple styles:
Full Badge
Shows label, grade, and score
Grade Only
Just the letter grade
Score Only
Numeric score out of 100
Grade Colors
Get Your Badge
Manual Badge URLs
If you know your model ID, construct the badge URL directly:
Static Badge (SVG)
https://hugginghugh.com/badges/{author}_{model}/badge.svg
Replace {author}_{model} with your model ID (slashes become underscores)
Dynamic Badge (Shields.io)
https://img.shields.io/endpoint?url=https://hugginghugh.com/api/badges/{author}_{model}.json
Updates automatically when your score changes
Report Link
https://hugginghugh.com/reports/{author}_{model}/
Full security report for your model
Example: meta-llama/Llama-2-7b
- Badge:
https://hugginghugh.com/badges/meta-llama_Llama-2-7b/badge.svg - JSON:
https://hugginghugh.com/api/badges/meta-llama_Llama-2-7b.json - Report:
https://hugginghugh.com/reports/meta-llama_Llama-2-7b/
FAQ
When do badges update?
Static SVG badges update daily during our scan (02:00 UTC). Shields.io dynamic badges update whenever shields.io re-fetches the JSON endpoint (typically within hours).
My model isn't found?
We currently scan the top 1000 most downloaded models on HuggingFace. If your model isn't in the top 1000, it won't have a badge yet. We're working on expanding coverage.
Can I improve my score?
Yes! The easiest improvements are: use SafeTensors format (not pickle), add clear license info, and keep dependencies updated. See our Trust Score Guide for details.
Is this official?
HuggingHugh is an independent community project. We are not affiliated with HuggingFace. Badges indicate our assessment based on publicly available metadata.