Skip to content
Security

Security model

Non-custodial by design. The protocol protects user funds with on-chain rules, and the backend is read-only with respect to wallets.

No custody, no private keys

The backend never holds a wallet key and never asks for a seed phrase. Buys, sells, claims, stakes, and referrals are signed by the user.

Permanent pool rules

Launch liquidity is a permanent single-sided v4 position. There is no graduation, migration, or later unlock path.

Fee isolation

The 1.30% fee and its platform, creator, referrer, and staker shares are enforced on-chain. External routers pay the same fee through hook deltas.

Bounded Uxp minting

Only authorized curves can mint Uxp. Daily minting is capped at 20,000,000 Uxp per wallet per UTC day; there is no global cap. External attribution is forced to the transaction origin, so router data cannot rotate wallets.

Data integrity

Database writes are idempotent. Candle and hourly aggregates use event-keyed triggers, and a single writer lease prevents duplicate background writers.

Backend controls

  • Parameterized queries only. The browser never talks to the database or a Base RPC directly.
  • Production database connections require TLS and scoped credentials. Insecure local URLs are rejected.
  • Creator social updates require an EIP-191 signature, a five minute timestamp, and a single-use nonce.
  • Rate limits cover HTTP, WebSocket subscriptions, nonces, and creator link edits.
  • Security headers, explicit CORS origins, request size limits, and protected metrics in production.
  • The write path retries transient failures without double counting volume.
  • One writer lease owns the indexer and reconciler. Read replicas can run without a second writer.
  • Referral codes are unique, random, and resolved server-side. Codes never expose private data.
  • Builder verification uses wallet-signed, single-use codes that expire in 30 minutes. The backend reads the README itself and never accepts a README body from the client.

Responsible disclosure

If you find a vulnerability, do not open a public issue or publish a proof of concept. Send the affected contract or endpoint, reproduction steps, impact, and a suggested fix. Please allow time for a fix before any public disclosure.