manav lite · agent authority
approved agents
only.
Hand an AI agent real authority - scoped, time-boxed, and revocable - always backed by a verified human. Any service can check it in a glance and turn away anything that isn’t approved.
Backed by a verified human · revoke anytime · nothing to leak
🧑 human-backed⏱ time-boxed🔌 revocable🔎 verifiable
see it in action
Stamp an agent 👇
Prove you’re human, then hand an AI agent named powers and an expiry. Out comes a credential any service can verify - and you can kill on the spot.
manav lite · agentscoped · revocable
Prove you’re human, then hand an agent scoped powers · the stamp it mints is a real, revocable credential.
approved agents only
Watch a service approve - or refuse.
This is exactly what a service sees when an agent knocks. Approve one, then revoke it and watch access die on the spot - no keys, no internals, just a yes or no.
service · access checkmanav verify
Runs against the agent you stamped above - live, on the real endpoint.
what it gates
Real powers. Real limits.
Give an agent exactly what it needs - and a hard, verifiable boundary it can’t cross.
how approval works
A human is always behind it.
🧑
1 · A real human proves it’s themOne glance, on-device. No account, no password - the identity is the person, not the device.
🔏
2 · They stamp the agentNamed scopes plus an expiry. The stamp inherits the human’s authority - and nothing more.
🔎
3 · Any service verifies & gatesValid, in-scope, unexpired → in. Revoke and it’s refused instantly, everywhere.
no internals exposed
A stamp, not your identity.
An agent carries a stamp - never your key, never your face. A service learns only three things: is it valid, what can it do, when does it expire. Nothing that reveals you, and nothing it could replay to impersonate you.
questions
How agent authority works.
What does Manav do for AI agents?
Manav lets a verified human stamp an AI agent with scoped, time-boxed, revocable authority. Any service can verify the stamp and refuse agents that are not approved.
How do you block unapproved agents?
Every service checks an agent's human-backed stamp before letting it act. No valid stamp means no access, so anonymous or rogue agents are turned away at the door.
Are agent actions tied to a human?
Yes. Every stamp names the verified human behind it, so each agent action traces back to an accountable person.
Can an agent's access be revoked?
Yes, instantly. Revoking a stamp turns the agent away everywhere on the spot, so a compromised or rogue agent is stopped immediately.
What can I scope an agent to do?
Named powers such as read inbox, send messages, book meetings, spend up to a limit, deploy code, or browse and buy, each with its own expiry.
Does a service see my identity?
No. A service sees only whether the stamp is valid, what it is scoped to do, and when it expires. Your key and your face are never exposed.
manav