FAQ

Questions, answered

Everything you need to know about the trust layer for the AI agent economy.

What is EVIDIQ?

EVIDIQ is the trust layer for the AI agent economy. Before an agent transacts with, delegates to, or relies on another agent or a paid service, EVIDIQ verifies its capability, scores its risk, checks its on-chain reputation, and returns a signed Trust Report.

How is a Trust Score calculated?

The score is deterministic and explainable — it combines four dimensions (identity, capability, reputation, risk) from the signals you supply and a live probe of the agent's endpoint. The same inputs always produce the same verdict, so it is auditable rather than a black box.

What makes the report tamper-evident?

Each paid report is hashed (keccak256), the full evidence is anchored on 0G mainnet Storage, and the verdict is signed with the EVIDIQ key (EIP-191). Anyone can re-fetch the evidence by its 0G root hash, re-hash it, and recover the signer.

Where does the AI analysis run?

The qualitative risk analysis runs on 0G Compute (model GLM-5.2) inside a TEE. The response records the on-chain provider address and request id, so the inference is attributable to a verified enclave — not a black-box API.

How do I connect it to my agent?

Add the remote MCP server in one line: `claude mcp add --transport http evidiq https://evidiq.dev/mcp`, or fetch the open skill with `curl -s https://evidiq.dev/skill.md`. See the Docs for details.

How much does it cost?

The skill and onboarding tools are free. The verify_agent trust check is metered with x402 (pay-per-call). Unpaid calls receive an HTTP 402 challenge; sign it and retry to run the check.

Which frameworks and chains does it work with?

Any agent framework — LangChain, AutoGen, CrewAI, LlamaIndex, Haystack, or custom. It settles on 0G (storage + compute/TEE) and X Layer (x402 payments), and lists on the OKX AI marketplace.

Does a high score guarantee a safe deal?

No. EVIDIQ produces evidence and a recommendation, not permission. It never holds funds or grants authority. For any deal with real exposure, pair the verdict with escrow or dispute rights.

Is EVIDIQ open?

Yes — the skill is served openly at /skill.md and the code is on GitHub. It complements agent-to-agent commerce skills such as Internet Court rather than replacing them.