EVIDIQ is live: a trust layer with proofs on 0G

Agents can already discover each other, negotiate, and pay. What they still lack is a way to know whether a counterparty they have never met can be trusted. EVIDIQ answers three questions before any value moves: can this agent do what it claims, how risky is the deal, and what is its standing — and can I prove the verdict later?

A verdict you can prove

Every check returns a Trust Report: a deterministic score across identity, capability, reputation, and risk, plus an explicit recommendation. The score is explainable by design — the same inputs always produce the same verdict, so it can be audited rather than trusted blindly.

The report is then made tamper-evident. The full evidence is anchored on 0G mainnet Storage (returning an on-chain transaction and a merkle root), a qualitative risk analysis runs on 0G Compute with GLM-5.2 inside a TEE (recording the provider address and request id), and the verdict is signed with the EVIDIQ key. Anyone can re-fetch the evidence by its root hash, re-hash it, and recover the signer.

Open, and easy to adopt

EVIDIQ ships the way agents actually consume capabilities: an open skill at /skill.md, a remote MCP server at /mcp, and an x402 pay-per-call endpoint so the core verify_agent check can be metered like any other agent service. Connect it to Claude Code, Cursor, or a custom agent in one line.

Evidence, not permission

EVIDIQ never holds funds or grants authority — it produces evidence and a recommendation. For any deal with real exposure, the report is meant to be paired with escrow or dispute rights. That is the point of a trust layer: make the risk legible, then let the parties choose their protection.