Stable Boy

Solana wallet stability score (0–100). Paste a wallet, get a receipt that explains exactly which behavioral signals dragged it up or down.

Live: stableboy.fun · X: @StableBoyPF · Token: $Stable

Stable Boy is fully open source under the MIT license. Self-host it, fork it, embed it, or contribute back.

Stack

• Next.js 15 (App Router) + TypeScript + Tailwind v4
• Supabase (Postgres) for permalinks + 24h score cache
• GMGN OpenAPI for wallet stats and funding source
• Bitquery EAP for pump.fun bundle co-buy detection
• Vercel for deploy

What ships in v1

Layer	Signal	Source	Status
1	Hold duration	GMGN wallet_stats.avg_holding_period	✓
1	Token diversity	GMGN wallet_stats.token_num	✓
1	Wallet age	GMGN common.created_at	✓
1	Win/loss distribution	GMGN PnL buckets	✓
1	PnL behavior	GMGN buy/sell ratio (proxy)	✓
1	Funding source	GMGN fund_from_address + Helius CEX labels	✓
1	SOL balance consistency	GMGN native_balance (snapshot)	✓ (proxy)
1	Paperhand ratio	FIFO entry/exit pairing on wallet_activity	✓
2	Bundle co-buy	Bitquery same-slot scan	✓
2	Sniper window	Bitquery launch-slot check	✓
2	Cluster proximity (1/2-hop)	Funder traversal × bundler_taglog	✓ (taglog grows from /api/accumulate-tags)
2	Funding daisy-chain	Funder chain freshness	✓
2	Wash trade	Round-trip pattern detector on activity	✓ (cross-wallet wash v3)
—	Trading activity heatmap	GMGN wallet_activity bucketed by day	✓

Setup

1. Install

npm install

2. Environment

Copy .env.example to .env.local and fill in:

• GMGN_API_KEY — read-only key from gmgn.ai/ai (Trading disabled is fine)
• BITQUERY_TOKEN — Developer plan access token from bitquery.io
• HELIUS_API_KEY — free-tier key from helius.dev. Optional: if unset, funder labels return null and funding source falls back to a v1 heuristic. Required for the +13 "direct from CEX" boost on legitimate wallets.
• NEXT_PUBLIC_SUPABASE_URL — from Supabase project settings → API
• SUPABASE_SERVICE_ROLE_KEY — from same page (server-side only, never expose to browser)
• NEXT_PUBLIC_SITE_URL — http://localhost:3000 for dev, your Vercel URL for prod

3. Database

In your Supabase project's SQL Editor, paste and run supabase/migrations/0001_init.sql.

This creates three tables:

• scored_wallets — receipts cached for 24h, keyed by wallet address
• funder_lookups — immutable record of who funded which wallet (v2 traversal)
• bundler_taglog — append-only log of per-token bundler/sniper/insider observations (v2)

4. Run

npm run dev

Open http://localhost:3000, paste a wallet, see a score.

5. Deploy

Push to GitHub, import the repo on Vercel, paste the same env vars in Vercel project settings, deploy.

Routes

• / — paste a wallet, get a score
• /w/<addr> — receipt page (cached 24h)
• /w/<addr>/badge.svg — embeddable SVG score badge
• /w/<addr>/opengraph-image — dynamic OG image for share previews
• /compare/<a>/<b> — side-by-side comparison with shared-funder detection
• /leaderboard — top 20 diamonds + bottom 20 bundlers, last 7 days
• /how — methodology page: factor weights, penalty triggers, sources, limitations

API

POST /api/score

{ "wallet": "7reR1o13S4VXpcTmB6Wtkrvi7wpba1C3x2LEmYEBRvLM", "force": false }

Returns { receipt: Receipt, cached: boolean }. If force is true, skips the cache.

GET /api/score?wallet=<address> — cache-only read, 404 if not yet scored.

GET /w/<address> — server-rendered receipt page, computes on demand if no cache.

Cost / rate limits

• GMGN free tier: 1 request/second. The client serializes calls through a promise chain so concurrent scorings queue rather than 429.
• Bitquery Developer plan: 1,000 points/month. Each score consumes ~30 points, so the free tier covers ~33 fresh scores per month. The 24h Supabase cache is what makes this practical at any user volume.

Score architecture

Two layers:

1. Behavioral score (0–100) — weighted average of normalized factors. Describes how the wallet trades.
2. Trust penalties (−60 cap) — hard deductions for cluster/bundler/insider patterns. Stack additively.

final = clamp(layer1 + layer2, 0, 100)

Verdict bands:

• 80–100 → diamond
• 50–79 → active trader
• 20–49 → degen / sniper
• 0–19 → bundler-adjacent

Wallets with <10 lifetime trades return insufficient_data rather than a fake score.

See src/lib/scoring/curves.ts for the per-factor curves and src/lib/scoring/weights.ts for the weight table.

Contributing

PRs welcome. Areas where help is especially useful:

• New Layer-2 signals. Wash trades across multiple wallets, dormant-then-coordinated, repeated bundle counterparts — anything that catches a pattern we currently miss.
• Score curve tuning. If you have a wallet you know is a bundler/diamond/etc. and the score disagrees, open an issue with the wallet address + what you think the score should be. The curves in src/lib/scoring/curves.ts are intentionally simple.
• Additional chains. Codebase is Solana-only today; the scoring engine is chain-agnostic and could extend to Base, BSC, etc. with new GMGN/Bitquery query backends.
• Performance. Fresh-score time is 8–12 seconds. The bottleneck is GMGN's 1 req/s rate limit. Better caching or parallelization could cut this.

Before submitting a PR:

• npm run typecheck passes
• npm run lint passes (when configured)
• No new env vars without a default that degrades gracefully

Acknowledgments

Data partners: GMGN, Bitquery, Helius. None of these companies sponsor or endorse Stable Boy — we're just paying customers / users of their APIs.

The aesthetic is inspired by Arkham Intelligence's information-dense crypto-forensics dashboards.

Extending

• Add Helius Wallet API for proper CEX-vs-daisy-chain funding classification → upgrade fundingSourceScore
• Implement wallet_activity parsing → unlock paperhand ratio + full-dump detection
• Implement recursive funder traversal (max 3 hops) → unlock cluster proximity penalties
• Implement token_top_traders per wallet token → unlock per-wallet sniper/bundler/insider tag accumulation in bundler_taglog