Permission infrastructure for autonomous agents

Give agents power.
Keep the leash.

Leash turns on-chain authority into a permission that can be checked, scoped, and revoked before every action.

Test the live gate ↓See how it works ↘
✓ ENSv2 on Sepolia ✓ ERC-8004 identity ✓ Live revocation
LIVE SYSTEMSEPOLIA · 11155111
Authority check
R
researcherresearcher.leash.eth
✓
requested capabilityinference.call
ENS roleROLE_SET_RESOLVER ✓
DECISIONALLOW
↗

Scoped to one capability · expires in 30s

Authority sourceENSv2Sepolia
Agent identityERC-8004Sepolia
Credential30 secsingle-use scope
Current statusSetup neededleash.eth
The problem with permanent keys

Agents are becoming capable.
Permission systems are not.

One leaked key can become unlimited access. One stale allowlist can outlive the person who created it. Leash makes authority explicit, short-lived, and observable.

01
✦

Authority on-chain

ENSv2 stores who an agent is and which capabilities it holds. No hidden allowlist, no guesswork.

02
✦

Scoped credentials

Each approved request receives a short-lived credential for exactly one capability.

03
✦

Revocation that works

Every protected call checks live authority again. Revoke once, and the next call fails.

One request, three checks
01IdentifyAgent + wallet
→
02AuthorizeENS role
→
03ScopeShort-lived credential
→
04Re-checkBefore execution
Live gate · try it yourself

Does this agent have permission right now?

Use the configured demo agent to read the real ENSv2 state on Sepolia. Change the role, revoke it, and watch the next request get denied.

01

Ask for access

READ-ONLY CHECK
Demo agent loaded ·
02

Get the answer

EXPLAINABLE RESULT
✦

Waiting for a request

Every result includes the verdict, reason code, block number, role bitmap, and credential scope.

Proof, not promises

Every decision leaves a trail.

Open full audit ↗

No decisions recorded yet.