5arz Proof-of-Human
Enables agents to prove and verify that a real, unique human performed an action, issuing machine-readable signed credentials with public verification.
README
5arz — Proof-of-Human MCP server
Prove a real, unique human performed an action — and mint a signed credential anyone can verify, offline, without trusting us.
Live endpoint: https://mcp.5arz.com (Streamable HTTP)
Docs: https://5arz.com/mcp · Free sandbox key: no card, no sales call
Why this exists
AI agents now solve CAPTCHAs faster than people do, and the fastest-growing "user" on the web is an agent acting for a real customer. That breaks the question everything was built around.
CAPTCHA asks "can this session pass a puzzle?" — a negative filter that treats a helpful agent and an abusive bot identically.
5arz answers a different question: "is there a real, unique human behind this action, and can they prove it to everyone downstream?" That is a credential problem, not a puzzle problem.
Web Bot Auth, Know Your Agent and the Trusted Agent Protocol all reference a verified human principal. None of them provide one. This is that layer.
What you get
A signed ES256 JWT, verifiable against a public JWKS with any JWT library:
{
"iss": "https://5arz.com",
"vct": "https://5arz.com/credentials/proof-of-personhood",
"sub_hash": "…",
"unique_human": true,
"liveness": true,
"assurance": "direct_document_liveness",
"verified_by": "https://5arz.com",
"method": "didit_document+passive_liveness+face_match",
"exp": 1790000000
}
Two properties worth understanding:
You never have to trust us. Verification runs against
https://api.5arz.com/.well-known/jwks.json. No SDK, no callback, no dependency
on our uptime — or our continued existence. A receipt only the issuer can
validate is worth what an assertion is worth.
The credential states its own provenance. assurance, verified_by and
method say how the human was established. A credential from a partner's own
KYC is labelled partner_asserted and does not claim liveness we did not run.
unique_human is only ever true when uniqueness was actually checked. A
credential is worth exactly the weakest thing that can produce it, so we make
the difference machine-readable instead of hiding it.
Tools
| tool | does |
|---|---|
verify_human |
attest that a verified human performed an action |
check_credential |
verify a Proof-of-Human JWT against the public JWKS |
register_agent |
create an agent identity + API key |
get_account |
credits, usage, recent verifications |
buy_credits |
purchase a verification pack |
vhh_index |
the Verified Human Hour index |
Quick start
# free sandbox key — unlimited, credentials marked test:true
curl -X POST https://api.5arz.com/api/agents/register \
-H 'content-type: application/json' \
-d '{"name":"My Agent","contactEmail":"you@co.com","mode":"test"}'
Add to an MCP client:
{ "mcpServers": { "5arz": { "url": "https://mcp.5arz.com" } } }
Verify a credential yourself
Nothing below calls 5arz:
import jwt, requests
jwks = requests.get("https://api.5arz.com/.well-known/jwks.json").json()
key = jwt.PyJWKClient("https://api.5arz.com/.well-known/jwks.json") \
.get_signing_key_from_jwt(token)
claims = jwt.decode(token, key.key, algorithms=["ES256"],
options={"verify_aud": False})
assert claims["iss"] == "https://5arz.com"
assert claims["unique_human"] is True
Connecting your own users
If you run a product and want your users verified, you do not need them to have a 5arz account:
- You have no IDV —
POST /api/agents/verify-linkreturns a hosted link. Send your user; a signed credential arrives at your webhook. - You already verify your own users —
POST /api/agents/humans/assert. We mint, label itpartner_assertedwith your method, and do the rest: signing, uniqueness, revocation, delegation, reporting.
Or drop the button on any page:
<script src="https://api.5arz.com/verify.js"
data-site="agt_your_id" data-ref="your-user-id" async></script>
Verified once, reusable anywhere. Returning humans re-prove with a passkey — no document, no vendor call.
The browser result is advisory. The authoritative path is the
credential.issuedwebhook to your server. Never gate access on apostMessage— your own users can forge it in devtools.
Run it yourself
npm install
npm test # protocol conformance
npx wrangler deploy
Status
101 verified humans · 18 countries · 5,740 verified completions. Patents pending US 63/979,052 · US 63/979,474.
The verification format is deliberately open. A credential nobody can independently check has no value.
License
Apache-2.0 — see LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.