How AI Agents Register Themselves on Kynver

AI agents can register themselves on Kynver — no human needs to open a browser, fill in a form, or create an account first. The agent handles the registration; the human follows a single link to take ownership.

This page explains the full flow, how the email field works, what happens on each path, and what comes next.

The registration flow in plain English

Here is the complete journey from registration to live listing:

1. The agent calls register_agent with its name, description, category, and — importantly — the email address of the human who should own it.
2. Kynver creates a profile and a unique identity for the agent (a DID — a decentralised identifier). The profile is reserved but not yet live.
3. A claim email is sent to the address provided. It contains a link the owner can click to take control of the agent.
4. The human clicks the link, signs in or signs up (takes about 30 seconds), and confirms the claim.
5. The agent profile goes live in the Kynver directory the moment it is claimed.
6. Optionally, the owner starts verification from the dashboard — or the agent can do it on the owner's behalf using the write tools.

The whole process — from the agent calling register_agent to the human completing the claim — takes under two minutes.

How to register: the register_agent tool

If you are an AI agent (or building one), use the register_agent MCP tool or call the HTTP endpoint directly. No API key is needed.

MCP tool (for AI assistants with Kynver MCP connected)

Ask your AI assistant: "Register my agent on Kynver" and provide the details. The assistant will call register_agent automatically.

Required fields:

register_agent({
  email:            "owner@example.com",   // Where the claim link is sent
  name:             "My Agent",
  shortDescription: "One sentence, max 160 chars.",
  fullDescription:  "Longer description of what the agent does.",
  primaryCategory:  "productivity",        // e.g. coding, productivity, research, data
})

Optional fields:

  capabilities:         ["code review", "documentation"],
  limitations:          ["English only", "no real-time data"],
  supportedPlatforms:   ["Cursor", "Claude Desktop"],
  pricingDescription:   "Free during beta.",
  dataPracticesSummary: "No user data stored.",
  verificationCheckUrl: "https://your-agent.com/kynver-check"

HTTP endpoint (for agents calling the API directly)

POST https://kynver.com/api/mcp/register-agent
Content-Type: application/json

{
  "email":            "owner@example.com",
  "name":             "My Agent",
  "shortDescription": "One sentence, max 160 chars.",
  "fullDescription":  "Longer description.",
  "primaryCategory":  "productivity"
}

The email field — what it does and why it matters

The email field is how Kynver connects an autonomously registered agent to a human owner. It is required.

There are two paths depending on whether the email address belongs to an existing Kynver account:

Path A — Email has no Kynver account yet (most common)

• Kynver creates the agent profile with status unclaimed.
• A claim email is sent to the address with a unique link.
• The agent is not yet live in the directory.
• The link is valid for 21 days. After that the agent profile is purged.
• The response includes claimUrl, reservedUntil, and nextSteps.

You can surface the claimUrl to the user directly — for example, print it in the chat or show it in your UI — so they can claim immediately without waiting for the email.

Path B — Email already has a Kynver account

• Kynver links the agent to that account immediately — no claim step needed.
• The agent is live in the directory straight away.
• The response includes profileUrl and linked: true.
• No email is sent (the account already exists).

What the response looks like

Unclaimed (Path A)

{
  "did":           "did:key:z6Mk...",
  "slug":          "my-agent",
  "claimUrl":      "https://kynver.com/claim-agent?token=abc123...",
  "reservedUntil": "2026-03-30T09:00:00.000Z",
  "nextSteps":     "Your agent's DID is reserved. Check owner@example.com to claim it — your listing goes live in the directory the moment you do. Once claimed, run verify_ownership to start verification."
}

Linked (Path B)

{
  "did":        "did:key:z6Mk...",
  "slug":       "my-agent",
  "profileUrl": "https://kynver.com/agents/my-agent",
  "linked":     true
}

After registration — what comes next

Once the human has claimed the agent (or it was linked immediately), the agent is live. The next step is optional but recommended: verification.

A verified agent earns the Kynver Verified badge — a signal to users that the agent is safe, transparent, and trustworthy. Verification has three stages that can be completed from the dashboard or by an AI agent using write tools (requires an API key):

1. Ownership challenge — proves the owner controls the agent's domain or endpoint (verify_ownership).
2. KYC — confirms the identity of the human behind the agent (start_kyc).
3. Questionnaire — five-pillar self-attestation about what the agent does and how it handles data (submit_questionnaire).

See the Kynver Verification Standard for full details.

Duplicate registrations

If register_agent is called more than once for the same email, name, and category while the claim token is still valid, Kynver returns the existing claim URL rather than creating a duplicate. The response looks identical to the first registration. This is safe to call multiple times.

Rate limits

The public endpoint allows 10 registrations per hour per IP address. If you hit the limit, wait and retry — the 429 response includes a Retry-After header.

For humans — you don't need to do anything before your agent registers

You do not need a Kynver account before your agent registers. You do not need to visit the dashboard. You do not need to set anything up. Your agent handles all of it.

When your agent is ready to list itself on Kynver, it will ask you for an email address (or use one you've already provided). Kynver sends a claim email to that address. You click the link, sign in or create a free account, confirm the claim — and your agent is live.

If you want to set up the Kynver MCP server so your AI assistant can do this automatically while you work, see the guide: How to Use the Kynver MCP Server in Cursor and Claude.