Give every agent its own .
The control plane that gives every agent its own credentials and audit trail — governed by signed mandates you control and can revoke in one step.
No shared secrets. · No loose scripts. · One control plane.
AI delegation didn’t get harder.
It got stuck.
Wrapper scripts work in a sandbox, but collapse once agents touch money or production systems. AgentTag replaces keys with governed identities, ensuring execution compounds safely instead of drifting.
How does a single control plane govern every agent?
AgentTag sits between your agents and the world. Every action flows through one policy engine, one identity layer, and one tamper‑evident ledger — whatever framework, tool, or cloud you use.
One MCP server. Eight tools.
Drop AgentTag into Claude Desktop or any MCP‑compatible agent. The agent sees only governed tools; every call goes through policy, vault, and audit, and no capability runs without a verdict.
The agent can request actions, but nothing completes unless policy allows it.
Everything you need to govern AI action.
Track, analyze, and authorize every request made by your autonomous agents across tools, hosts, and networks in real time.
Every node in any cloud
Govern execution across private microVMs, secure credential enclaves, and KMS-backed environments.
Govern any agent
Apply the same rules across CrewAI, LangChain, Claude Desktop, and any MCP or A2A client.
Connect your tools
Bind third-party credentials behind secure, sandboxed API surfaces.
Biometric consent
Fails closed on timeouts. Authorizations require on-device biometric check signed by your key.
Durable mandates ledger
Every decision is logged and cryptographically verifiable.
Interrupted only when it genuinely matters.
Routine work runs untouched. When the agent reaches a new payee, a spend over your threshold, or anything irreversible, it pauses and asks — and your approval is signed on-device, never a blank cheque.
AgentTagSTEP-UPAgent Research wants to pay $840.00 to Acme Data Inc
category: data · new merchant · mnd_01H…
Signed with your passkey · on-device
How does the audit ledger secure agent actions?
A tamper‑evident ledger records every request, decision, and approval. Hash‑chaining makes quiet rewrites impossible.
Why choose delegated identity over shared keys?
Password managers share secrets. DIY scripts skip governance. AgentTag is identity-first, MCP-native, and accountable by construction.
| Cryptographic agent identity & delegation | AliasKit | anima | AgentWallet | AgentTag |
|---|---|---|---|---|
| DID-anchored agent passports | ✕ | ✕ | ✕ | ✓ |
| Cryptographic, attenuable mandates | ✕ | ✕ | ✕ | ✓ |
| Real-world primitives (pay, comms) | ✓ | ✓ | ✓ | ✓ |
| On-device passkey approvals | ✕ | ✕ | ✕ | ✓ |
| Hash-chained, offline-verifiable ledger | ✕ | ✕ | ✕ | ✓ |
Free while we’re in beta.
All primitives, policy, and the hash‑chained ledger are unlocked for public beta. No card. No seats. No hidden catch.
Bring your first agent online and govern it end to end.
- All five primitives, live
- Multiple agent passports
- Full policy engine and step-up
- Passkey approvals on-device
- Hash-chained audit ledger
- Direct line to the founders
No credit card. Cancel anytime. Your data stays yours.
When we reach general availability, you’ll pay for autonomy, not seats.
- 30 days’ notice before any new plan starts
- A generous free tier that stays free
- Founding-user pricing locked in
- Enterprise SSO, SLA, and on-prem support when needed
Questions, answered.
Everything you need to know about giving an agent its own governed identity.
Give your agent a passport.
Bring your first agent online with governed identity, signed permissions, and a verifiable audit trail.