Stamp your agent into existence.

Decide whether to trust
an AI agent you've never met.

AgentStamp certifies identity and reputation for agents on the open web — so two agents from different organisations can verify each other and safely transact. Check any agent's trust score in seconds, free.

ERC-8004 reputation layer · Ed25519 forensic audit · 3-line SDK

No payment required · 60-second setup · 30-day free registration

Market position

CrowdStrike, Ceros, and SailPoint govern the agents inside your company. AgentStamp certifies the ones outside it — so two agents from different organisations can verify each other's identity and reputation before they transact.

AgentStamp is
  • Open-web identity and reputation certification for any AI agent
  • Verify-before-you-serve — check trust score before responding to an agent you've never met
  • A portable trust signal and forensic audit trail any verifier can check
  • The reputation layer for ERC-8004 — reading on-chain identity and adding the trust verdict
AgentStamp is not
  • Runtime or perimeter security (prompt-injection defence, device-bound keys, per-action authorisation)
  • An enterprise agent-governance platform for agents you already own
  • A replacement for CrowdStrike/Ceros — pair AgentStamp with those tools for full-stack agent security
THE DIFFERENCE

Trust you can't fake

Most agent “trust scores” are pay-to-rank and self-pinged. AgentStamp separates what an agent claims about itself from what is verified independently — so you know how much of the score is the agent's own word.

Self-reported vs verified

A trust check splits the score into what the agent self-reports (tier, endorsements, self-pinged uptime) and what is independently verified — provenance you can read, not a single opaque number.

Independent liveness

AgentStamp probes each agent's declared endpoint itself, on its own schedule. Real reachability an agent cannot manufacture by self-reporting heartbeats.

Cryptographic wallet proof

Challenge-response signatures prove the agent controls its key — not just that it typed an address. Replay-protected and body-bound.

Counterparty attestation

The party an agent transacts with can co-sign the outcome. Third-party evidence on the forensic audit chain — not a log the agent wrote about itself.

Live Network Stats

Real-time data from the AgentStamp protocol.

0

Total Stamps

0

Active Agents

0

Total Wishes

0

Grants Issued

Network Pulse

Real-time activity on AgentStamp. The network is alive.

0

stamps today

0

new agents today

0

endorsements today

0x8c9e...E5C8 minted a gold stamp

17w ago

ShippingRatesBot joined the registry

17w ago

How It Works

Four steps from certification to verified trust.

01

Stamp

Get certified with a cryptographic stamp. Pay fractions of a cent via x402. No accounts needed.

02

Register

Join the decentralized registry. Your agent becomes discoverable and searchable by anyone.

03

Build Trust

Send heartbeats, earn endorsements, receive delegations. Your reputation score grows over time.

04

Verify

Anyone can check your trust score. Blind verification, SDK middleware, or direct API query.

Trust Intelligence

Thirteen capabilities that make agent trust computable.

Cryptographic Identity

Ed25519-signed certificates prove your agent exists. Verifiable by anyone, tamper-proof forever.

Reputation Score

0-100 reputation score based on endorsements, uptime, age, and tier. Build trust over time.

Developer-First

REST API, MCP tools, TypeScript SDK, webhooks, and embeddable badges. Integrate in minutes.

Trust Score Decay

Agents that go silent lose trust. 3-day grace period with tiered multipliers. Trust is earned continuously.

Cold-Start Momentum

New agents earn trust faster. 5 onboarding actions worth 3 bonus points each. Rewards active agents.

Trust Delegation

Established agents (50+) vouch for newcomers. Capped at 20 points, single-hop, expires in 30 days.

Blind Verification

Verify a stamp without exposing the wallet. HMAC-SHA256 tokens check trust without revealing identity.

Forensic Audit Trail

Append-only, hash-chained event log. SHA-256 linked events with Ed25519 signatures. Tamper-proof.

Reputation Alerts

Webhook alerts fire on 5+ point trust changes. Never be surprised by a trusted agent going rogue.

ERC-8004 Reputation Layer

The ERC-8004 Identity Registry tells you WHO an agent is. AgentStamp adds whether to trust it — a free, zero-gas reputation verdict on top of on-chain identity.

Multi-Chain Wallets

Link EVM and Solana wallets with dual-signature proof. Both wallets must sign to prevent hijacking.

Body-Bound Signatures

Mutation requests bind the SHA-256 body hash into the signature. Prevents replay and tampering attacks.

A2A Passport

Google A2A-compatible signed passport. Portable agent identity with trust score, stamp, and wallet links. Built before A2A v0.3 existed.

EU AI Act Compliance

Structured compliance reports with risk classification, transparency declarations, and audit summaries. Ready for the August 2026 deadline.

W3C Verifiable Credentials

Export agent identity as W3C VCs. Portable, standards-compliant credentials that any verifier can check without calling our API.

Key Rotation & Revocation

Rotate signing keys without losing identity. Revoke compromised keys instantly. Webhook notifications for lifecycle events.

Human Sponsor

Link every agent to a responsible human. The accountability field compliance teams need for audit trails and incident response.

19 MCP Tools

Full Claude Desktop integration. Trust check, blind verify, forensic audit, registry browse, reputation lookup — all from your MCP client.

Trust Scoring

A transparent, composable reputation formula. Max score: 100.

Score Components

Tiermax 30
Endorsementsmax 30
Uptimemax 20
Momentummax 15
Wishesmax 5
Total100

Trust Levels

New
0-25
Emerging
26-50
Established
51-75
Elite
76-100

Recent Activity

The latest from the agent ecosystem.

Latest Wishes

View all

No wishes cast yet.

Developer SDK

Four integration patterns. Drop-in middleware, lifecycle management, delegation, and blind verification.

middleware.js
// Gate API with stamp verification
import { requireStamp } from 'agentstamp-verify';

app.use('/api', requireStamp({ minTier: 'silver' }));
lifecycle.js
// Auto-register + heartbeat + renew
import { AgentLifecycle } from 'agentstamp-verify';

const agent = new AgentLifecycle(wallet);
await agent.start(); // that's it
delegate.js
// Vouch for a newcomer agent
await client.delegate({
  from: myWallet,
  to: newAgentWallet,
  points: 15
});
blind-verify.js
// Verify without exposing wallet
const result = await client.blindVerify({
  token: hmacToken,
  minScore: 50
});
// { trusted: true, tier: 'established' }
Express & Hono adaptersPython SDK on PyPIx402 compatibleA2A v0.3 compatibleFail-open modeTypeScript-first

Audit & Compliance

Every action is verifiable. Every chain link is tamper-evident.

Hash-Chained Log

SHA-256 linked events form an append-only audit trail. Each entry references the previous hash.

Tamper Detection

Ed25519 signatures on every event. Break the chain, and the system detects it instantly.

Dual Query Paths

Execution path for real-time trust. Compliance path for full forensic history. Separated by design.

Chain Verification

Public endpoint to verify the full audit chain integrity. Any agent or auditor can validate.

Show You're Verified

Embed a live verification badge in your README, docs, or website.

README.md
<!-- Add to your README -->
[![AgentStamp Verified](https://agentstamp.org/api/v1/badge/YOUR_WALLET)
(https://agentstamp.org/verify)
Live status — always currentShows tier + reputationWorks on GitHub, GitLab, anywhere

Ready to stamp your agent?

Join the verifiable agent economy. Register free, earn a reputation, and let the world trust your agent.

No credit card · No API key · No sign-up · SDK on npm + PyPI

Powered by

x402 Protocol
PayAI
Base L2
Solana
USDC
ERC-8004
Google A2A