H-Series H-Index H-Seal H-Grant H-Relay H-Scope H-Pact H-Gate H-Cert H-Agent Contact
H-Series

For everyone

Give your AI agent an identity

An agent that is verifiably yours, with rules it must follow, on any AI platform. You do not need to be a developer for most of this. If you can use a browser and install an app, you can do the main steps in about 15 minutes.

An identity is a key

An agent's identity is not a name or a username. It is a key, like a keycard or a passport the agent carries. Whoever holds the key is that identity. From the key comes a public address you can share; the secret key is the part you protect. Because the identity is a key and not a platform account, it is portable: the same agent identity works whether the agent's brain is Claude today and Gemini tomorrow.

  1. 1

    You, the owner. This agent is mine.

  2. 2

    The agent's identity, a key it holds.

  3. 3

    The rules: may do X, up to a limit, until a date.

Three things, in order: your identity, the agent's identity, and the rules that connect them.

The five steps

1

Set up your own identity

This is the identity that says these agents belong to you. Connect a wallet, or just log in with Google. Use the same one every time so all your agents sit under one roster.

2

Create your agent's identity

Your agent needs its own key, separate from yours. In a wallet app, create a new account. Its address is your agent's identity; its secret key is what makes it real. Keep that secret safe; you give it to the agent in the last step.

3

Put your agent on your roster

This creates a public, tamper-proof record that the agent is yours. Paste the agent's address, give it a name, choose operated-by, and confirm. You only ever use its public address here, never its secret key.

4

Set the rules

A roster says who the agent is. A rule says what it may do. Pick your agent (straight from your roster), set what it may do, a spending cap, and an expiry. The limit is checked every time it acts, not left to the agent's good behavior.

5

Give the agent its key so it can act

The one technical step. The agent's code needs its secret key so it can prove its identity when it works. Store it where the agent runs, never in a prompt or a chat. Not technical? Hand this step to a developer, or use a template that already knows how to sign.

It works on every platform

Notice what was never mentioned in the steps: which AI you used. That is the point. The identity is the key, not the platform. The same identity, roster entry, and rules apply whether the agent's reasoning comes from Claude, ChatGPT, Gemini, Grok, or DeepSeek, and whether you move it from one to another. You are not renting an identity from a platform that can change the terms. You own it.

Staying safe

  • The secret key is everything. Anyone who has it can act as the agent. Never paste it into a chat or a prompt.
  • Caps and expiry are your seatbelt. Always set them. They protect you if the agent goes wrong or is taken over.
  • You can revoke. If something feels off, remove the agent's permission and it loses access.
  • One identity per agent. Separate keys mean a problem with one never touches the others.
  • Start small. A new agent gets a tiny cap and a short expiry; widen it once you trust it.

Ready to start

Put an agent on your roster, then set its rules.

Where this is today: the owner-facing parts (your identity, the roster, the rules) are point and click now. Creating the agent's key uses an ordinary wallet app. The final step, wiring that key into the agent so it can sign its own actions, still benefits from a developer or a ready-made template. Simpler one-click versions of that last step are coming. The foundation, an identity you own that works across every platform, is here now.