Veritrust ANS gives agents human-readable names and signed “cards”.
The same host also operates the MCP Trust Registry,
exposing a machine-readable catalogue of MCP servers at
/mcp for AI platforms and hosts.
ANS & MCP registry entries are ultimately verified against
did:web:veritrust.vc and the Veritrust Trust Policy.
/mcp
ANS focuses on human-readable naming and signed discovery for agents. The MCP registry focuses on MCP servers: where they live, what they do, and which credentials and issuers are trusted.
Global discovery metadata is also advertised via
https://veritrust.vc/.well-known/mcp-trust-registry.json.
ANS lets you resolve human-readable names like
ans.veritrust or agent.robert-cane.neksus
into signed “agent cards” that describe identity, capabilities, endpoints and status.
Clients call:
GET /resolve?name=<logical-name>
The response contains a JSON “card” plus a JWS signature. The
signature is verified against the ANS key published in the Veritrust DID document:
https://veritrust.vc/.well-known/did.json.
This makes “who runs this agent” and “what it can do” verifiable, not just documented.
{
"card": {
"name": "issuer.veritrust",
"version": "1.0.0",
"provider": "Veritrust",
"did": "did:web:veritrust.vc",
"capabilities": [
"vc-issuer",
"oidc4vci",
"status-list-2021"
],
"endpoints": [
{
"protocol": "https",
"url": "https://veritrust.vc/issuer/",
"purpose": "vc-issuer"
}
],
"status": "active",
"issued_at": "..."
},
"jws": "<compact signature>",
"kid": "did:web:veritrust.vc#ans-key-1"
}
name=ans.veritrustname=issuer.veritrustname=zone.neksus.ainame=agent.robert-cane.neksusANS helps organizations publish verifiable, protocol-agnostic metadata for agents, improving discovery, governance, and security across agent ecosystems. It aligns with work such as:
In addition to ANS, this host exposes the MCP Trust Registry API at /mcp.
It implements the MCP Trust Framework (MCPF) registry interface so that AI agents and platforms
can discover and verify MCP servers before using them.
GET /mcp/health – basic health checkGET /mcp/servers – list MCP serversGET /mcp/servers/{did} – single server by DIDGET /mcp/search – search by metadataGET /mcp/issuers – trusted issuersGET /mcp/revocations – revoked servers/credentials{
"did": "did:web:veritrust.vc:mcp:edgeguard-siem",
"endpoint": "https://edgeguard.cyberfort.lv/mcp",
"manifest": "https://edgeguard.cyberfort.lv/mcp/manifest.json",
"credentials": [
"https://veritrust.vc/status/agents-2025/edgeguard-mcp.json"
],
"metadata": {
"capabilities": ["telemetry","alerts"],
"organization": "Cyberfort",
"country": "LV",
"tags": ["security","siem"],
"status": "active"
}
}
MCP clients can start at:
https://veritrust.vc/.well-known/mcp-trust-registry.json
That document advertises ans.veritrust.vc/mcp as the registry base URL,
as well as the Veritrust issuer DID and key material used to issue MCPServerCredential verifiable credentials.
Hosts then query this registry to apply policy (“which MCP servers are allowed”) before enabling tools.
Use ANS for human-readable names and signed agent metadata. Use the MCP registry for machine-readable MCP server entries. Both can be integrated with your own policy engine and identity stack.
# Resolve the ANS root
curl "https://ans.veritrust.vc/resolve?name=ans.veritrust"
# Resolve an issuer
curl "https://ans.veritrust.vc/resolve?name=issuer.veritrust"
# List MCP servers
curl "https://ans.veritrust.vc/mcp/servers"
# Search for servers with a capability
curl "https://ans.veritrust.vc/mcp/search?capability=telemetry"
The MCP registry on this host follows the MCP Trust Framework (MCPF), which defines how registry entries, credentials and trust checks work.
github.com/Veritrust-VC/mcp-trust-framework →For a more complete explanation of the MCP Trust Framework, its relationship to MCP, and how registries and credentials fit together, see:
Veritrust MCP Trust Framework landing page →
ANS and the MCP registry live alongside Veritrust's issuer, status lists, DIDComm mediator and
other components, all anchored in did:web:veritrust.vc and the Veritrust Trust Policy.
For integration, security reviews or dedicated deployments, contact: [email protected]