HuggiMon — your AI trainer card from your Hugging Face profile

On Hugging Face, your public profile already tells a story: models you ship, Spaces you maintain, datasets you share, likes and downloads that add up. HuggiMon (huggimon.co) takes those signals — public data only — and turns them into a Pokémon TCG-style trainer card: 3D tilt, holo shaders, a follower binder, and a dedicated page per username.

Why I built it

The Hub is great at hosting ML, less at showing “who you are” at a glance. READMEs list repos; stats are split across models, datasets, and spaces. I wanted a playful summary — not a serious leaderboard, but a card you can open at huggimon.co/your-username, share on X, or drop in your Hub README.

• No auth: a username is enough; everything is read via the public Hugging Face API.
• Interactive card with 14 holo tiers based on your HF level.
• 3×3 binder: your followers become mini-cards to flip through.
• Downloadable PNG + Markdown snippet for your README.

How it works

The flow is linear: fetch → score → render. You open /{username}; HuggiMon pulls the user overview and public repos (models, datasets, spaces). A scoring module computes six stats (0–100), infers type, rarity, level, attacks, and energy (from likes). Client rendering uses pokemon-cards-css holo shaders; the server composes the card face as a 660×921 PNG.

/{username} → hf-fetcher → scoring → PokemonCard (holo + tilt)
                      ↳ /api/card/{username}/face → shareable PNG

The six stats

• MODEL — published models, likes, and downloads.
• DATA — datasets and their traction.
• SPACE — published Spaces and associated likes.
• IMPACT — aggregate likes + downloads across public work.
• COMMUNITY — followers and discussions.
• DOCS — share of repos with a description.

Holo, energy, and binder

Your HF level picks a holo tier among 14 (reverse holo → Secret Gold). Likes on your work become energy on the card. Your followers fill a paginated 3×3 binder with a page-flip animation — click any mini-card to inspect it enlarged.

Share and embed

Each profile has a public URL and a ready-to-paste README snippet:

[![HuggiMon](https://huggimon.co/api/card/ImTamsi/face)](https://huggimon.co/ImTamsi)

• GET /{username} — profile page (card + binder + share).
• GET /api/card/{username} — JSON metadata (stats, type, rarity, energy).
• GET /api/card/{username}/face — composed 660×921 PNG.
• GET /api/binder/{username} — follower binder page (?page= optional).

Tech stack

• Next.js — app in web/, deployed on Vercel (huggimon.co).
• pokemon-cards-css — 3D holo effects (tilt, glare, rarity shaders), React port via @react-spring/web.
• gitfut — inspiration for the /{username} pattern and README embed (applied to the HF Hub instead of GitHub).
• MIT — open source at github.com/Tamsi/huggimon.

Try it

Go to huggimon.co, enter an HF @username, or open huggimon.co/ImTamsi directly. It’s a Build Small Hackathon side project: fun first, but scoring is transparent — the repo is on GitHub if you want to challenge the formula or contribute.

• Site: huggimon.co
• Repo: github.com/Tamsi/huggimon