Verifier Getting Started

Who Are Verifiers

Verifiers are the referees of deals — when transaction parties disagree on results, they provide objective, impartial rulings that take effect directly on-chain.

The form of the verifier doesn't matter; what matters is: the ruling must be correct, timely, and tamper-proof.

Verifier income comes from verification fees — automatically settled by the contract after each successful result submission.

Three Implementation Forms

FormApplicable ScenariosDescription
Pure code programAPI-queryable, rule-based verification tasksPreferred approach. Highest determinism, no subjective judgment bias. For example, calling the Twitter API to check if a tweet exists, querying on-chain state, etc. Low operating costs, fast response, fully reproducible results.
Human verifierTasks requiring offline verification or subjective judgmentSuitable for scenarios that pure code cannot cover — such as physical delivery confirmation, content quality assessment, etc. Human verifiers verify off-chain and manually submit results on-chain.
AI AgentIn between the two, requiring some reasoning abilityCan handle tasks more complex than pure code, but less stable than pure code. Watch out for hallucination, prompt injection, and other risks.

Selection advice: Prefer pure code first. Use human verification for scenarios pure code cannot cover. AI Agents can be used but require thorough validation of stability and fairness.


Authority: What Makes People Trust You

No credentials, no platform endorsement. Authority comes entirely from on-chain historical records:

  • Traders check your success_count, fail_count, and timeout count before selecting you
  • Long-term high accuracy → higher search rankings → more quote requests
  • One act of misconduct = reputation reset to zero; on-chain records are permanent

Authority is not granted — it is earned. Early participants who consistently respond correctly possess competitive advantages that latecomers cannot replicate.


Reliability: How to Ensure Response Dependability

Consequences of timeout without responding: verification fee refunded, timeout recorded on-chain, credit damage. Countermeasures:

  • Monitoring & alerts — Off-chain services must have health checks configured to ensure timely notification on failure rather than silent failure
  • Redundancy design — Critical dependencies should have backup plans, ensuring a definitive conclusion within the deadline
  • Know your limits — If you cannot complete within the deadline, return accepted: false at the request_sign stage
  • Composite architecture — Multiple instances voting in parallel (majority rules) or sequential backup (switch on timeout), exposing only a single address externally

Code of Conduct

Verification parameters must be read from on-chain sources; never trust off-chain data. A signature is a commitment — once on-chain, it cannot be revoked. The verifier's duty is to report facts truthfully — make every effort to reach a definitive conclusion, and return 0 when unable to determine. Submitting conclusions without thorough verification constitutes a violation.