Skip to content

Accord

General-purpose, Schelling-point-based decentralized arbitration on Solana.

Any Solana program can file a dispute via a single CPI call. The Accord draws stake-weighted jurors using a verifiable random function, collects commit-reveal votes, and emits a ruling the filer reads lazily. No central authority picks judges — game-theoretic incentives make honesty the profitable strategy.

Why use it

  • Two CPI calls to integrate: create_dispute() → get_ruling(). The Accord has zero knowledge of your program's domain.
  • Schelling-point security: jurors vote truthfully because voting with the majority is the profitable strategy. Proven by Kleros (Ethereum, since 2019, 1000+ disputes).
  • Oracle-verified randomness: the magicblock VRF (RFC 9381) determines juror selection. Unbiasable, unpredictable, on-chain verifiable.
  • Five fraud predicates: duplicate, omission, wrong-stake, inflation, and not-sorted — every snapshot manipulation is detectable and punishable.
  • Exponential appeals: 3 → 7 → 15 → 31 jurors. Bribery becomes prohibitively expensive.

How it works

Your Program (Arbitrable)
    │
    ├─ create_dispute(subaccord, options, evidence_hash, fee)
    │      └─ Accord draws N jurors (VRF, stake-weighted)
    │      └─ Jurors commit hash(vote, salt) → reveal {vote, salt}
    │      └─ Majority → Ruling
    │
    └─ get_ruling(dispute_id) → winning_option

What you need

  1. A Subaccord — a specialized juror pool with its own staking token, min stake, and fee schedule. Use an existing one or create your own (permissionless).
  2. Capital — the filer pays jurors_per_dispute × fee_per_juror per round. Appeals cost exponentially more.
  3. The SDK — @accord/sdk for TypeScript, or raw Anchor IDL for any language.

Next steps

Deep reads