About this skill

What problem it solves

Autonomous AI agents often execute high-stakes actions while leaving only internal logs. Those logs can be edited and are weak evidence for AI Act compliance, legal accountability, or user trust. xProof targets proof-before-action: before an agent initiates a trade, publishes content, or runs a decision, it fixes the reasoning (WHY) and planned action (WHAT) as publicly verifiable on-chain proofs, turning opaque execution into an auditable workflow.

How the skill works

The core mechanism is privacy-first hash anchoring: the agent computes a SHA-256 digest locally and anchors only that digest on MultiversX, keeping raw content local. It supports x402 pay-per-call flows, usually without an API key, account, or human approval, and exposes an MCP JSON-RPC 2.0 endpoint with tools such as certify_file, audit_agent_session, investigate_proof, and register_trial. After a proof is generated, the agent or reviewer can query a 4W trail covering Who, What, When, and Why, then verify it publicly at xproof.app/proof/{id}.

Boundaries and notes

It fits trading or finance agents, research agents, content-generation agents, and enterprise compliance gates that need decision trails. It is not a general monitoring dashboard, and it does not replace access control, key management, business risk controls, or human approval. When integrating it, confirm the x402 payment path, MCP endpoint, trial quota, and target chain, and avoid putting sensitive raw content into fields that are not designed for it.

Screenshots

Use Cases

A trading agent records its risk check and order intent as an on-chain proof before execution.
A research agent certifies key evidence files with `certify_file` before publishing findings.
A content agent anchors AI-generation metadata before publication for compliance review.
An enterprise agent audits a session with `audit_agent_session` before contract or payment actions.

Best For

Quant engineers who need risk-check evidence before a trading agent executes.
Research-agent developers who need verifiable evidence before publishing findings.
Content-platform engineers who must record AI-generation metadata before publication.
Compliance engineers who need pre-action audit trails for enterprise agents.