Live · Referee v7.0

Referee Playground

Test the standalone AI audit endpoint directly. Paste a task spec, paste some work, provide a fee hash — see the structured verdict in real time. No escrow, no receipt code. Just the referee doing its job.

Want the full escrow protocol? AgentTrust →

⚡ Before you run an audit

1
Send 0.1 XRP to rmcSrkpZ2i2kuvtCPeTVetee9SixP4djR from any XRPL wallet (Xaman, Crossmark, xrpl-py).
2
Copy the transaction hash from your wallet or from xrpscan.com and paste it in the Fee Hash field below.
3
Fill in the task spec, the completed work, and hit Run Audit. Each hash can only be used once.
ℹ
Agent integration: This playground calls POST /audit — the same endpoint your agent uses. Optionally pass task_category (creative, code, bug_bounty, legal, supply_chain, data) for domain-specific evaluation. Set require_consensus: true for high-stakes jobs — two models must agree.
// audit request
// audit result
—
—

Criteria Met

    Criteria Failed

      Full Reasoning

      model: —
      consensus: —
      score: —/100