Consequences require evidence.

Your agent can act. Can it prove why it acted?

Make AI agents prove before they act: when an agent moves money, changes production, signs a contract, or delegates, commit a declared decision basis before execution and link a recorded result for independent verification. PBA verifies the published commitment and linked record—not private intention or arbitrary off-chain execution—and does not block actions at runtime. The operator must enforce its gate.

• Payment approval
• Production deploy
• Legal commitment
• Agent delegation

Integrate your agent — 10 free proofs, no wallet, no credit card. Or verify a public proof without an account.

Prove Before Act is the public accountability pattern. xProof is the reference implementation. Current live rate after the free proofs: $0.01 per certification — see /api/pricing.

Connect an agent with the MCP Server; inspect MCP discovery or start with the agent integration guide.

Why integrate before anyone requires it?

Evidence cannot be created after the fact. The right time to build an accountability record is before a client, auditor, insurer, or regulator asks for one.

1. The downside is asymmetric

   A first integration is small and predictable: 10 free proofs, then the current live rate is $0.01 per proof. Waiting costs nothing until someone asks for evidence that was never created. At that point, it cannot be reconstructed.

2. The proof must precede the incident

   A record written after a dispute shows what you say happened. An independently verifiable commitment shows what was declared before execution, not private intention or that an off-chain action occurred. The operator must enforce any action gate.

3. A verifiable history compounds

   Operators that start now can build a durable record of declared commitments and linked result records. That evidence can support review, but does not establish legal or regulatory compliance.

Integrate your agent · Inspect the Prove Before Act standard

How pre-execution evidence works

Build a verifiable trail around a declared decision basis and linked recorded result. A linked result is evidence of a record, not independent proof of arbitrary off-chain execution or the truth of its claims.

1. Declare a decision basis and intended action

   Your agent records the decision, declared justification, context, and intended action locally. This is not a request for internal chain-of-thought.

2. Hash locally

   A SHA-256 hash is computed locally from that declared record, a model output, data snapshot, or build artifact. Only the hash is transmitted — source material stays private.

3. Anchor before acting

   Use the proof_id to check independent confirmation of the commitment before acting, then link the outcome to that proof.

One principle. Two ways in.

Commit before the action. Integrate that rule into your agent, or inspect an existing public proof.

Simple pricing - One price. No subscription.

Current live rate: $0.01 per certification (see /api/pricing). Pay only for what you use. No hidden fees, no commitment.

• Unlimited certifications
• Public proof_id and blockchain transaction URL
• Public verification page
• Optional PDF export with QR shortcut
• MultiversX blockchain

Frequently asked questions

Is my file uploaded to your servers?

No. Only the SHA-256 hash is transmitted. Source data never leaves the agent's runtime environment.

What is the MultiversX blockchain?

MultiversX is a high-performance, eco-friendly European blockchain. Unlike Bitcoin, it consumes very little energy.

Does it have legal value?

A blockchain record can help show that a matching hash was anchored by a recorded time. Legal effect depends on the facts and jurisdiction; Prove Before Act is not legal advice or a compliance certification.

Make critical actions accountable

Operators can require independently verifiable pre-execution evidence before deployment, handoff, payment, or another external action. Current live rate: $0.01 per certification; see /api/pricing.