The Trust Layer for AI Agents
Give your agents an identity. Delegate permissions. Verify trust. Stop agent fraud before it starts.
AI Agents Are Everywhere. Trust Is Nowhere.
Agent Fraud
Any process can claim to be an agent. There's no standard way to verify it's who it says it is.
Agent-to-Agent Trust
When two agents need to collaborate, neither can prove its identity or capabilities to the other.
Delegation Verification
How does a service know an agent actually has permission from its owner? Today, it can't.
Scope Control
Agents need allowances — bounded permissions that limit what they can do. API keys don't cut it.
Agent Trust in 5 Lines of Code
Create an agent identity, delegate permissions with scoped allowances, and verify trust — all from a single, composable SDK.
import { createAgent, delegate, verifyPresentation } from "@credat/sdk"
// 1. Create an agent identity
const agent = await createAgent({
domain: "agent.acme.com"
})
// 2. Delegate scoped permissions
const delegation = await delegate({
agent: agent.did,
owner: ownerDid,
scopes: ["book:travel", "spend:500usd"]
})
// 3. Verify trust
const result = await verifyPresentation(presentation)
console.log(result.trusted) // trueWhat You Can Build
Travel Booking Agent
Your agent books flights and hotels on your behalf. Services verify it's authorized to spend up to your set limit — no API keys, no shared passwords.
Financial Agent with Spending Limits
Delegate a spending allowance to your agent. It can pay invoices, transfer funds, or manage subscriptions — all within the scope you defined.
Multi-Agent Workflows
Chain agents together where each one proves its identity and permissions to the next. Build complex pipelines where trust flows automatically.
Agent Platform Security
Run a marketplace where third-party agents interact with your APIs. Verify every agent's identity and scope before granting access.
Personal Assistant Agent
Your personal agent manages your calendar, emails, and tasks. It proves to every service that you authorized it — with exactly the right permissions.
Create an Agent
Give your agent a verifiable identity tied to your domain. One function call, no configuration.
Delegate Permissions
Define exactly what the agent can do with scoped allowances. Book travel, spend $500, read emails — you decide.
Challenge the Agent
When your agent arrives at a service, the service issues a challenge. The agent proves its identity and permissions.
Verify Trust
The service verifies the agent's identity, its owner's delegation, and the scopes — all in one call.
Create an Agent
Give your agent a verifiable identity tied to your domain. One function call, no configuration.
Delegate Permissions
Define exactly what the agent can do with scoped allowances. Book travel, spend $500, read emails — you decide.
Challenge the Agent
When your agent arrives at a service, the service issues a challenge. The agent proves its identity and permissions.
Verify Trust
The service verifies the agent's identity, its owner's delegation, and the scopes — all in one call.
Everything You Need to Trust Agents
Agent Identity
Create portable, verifiable identities for AI agents. Each agent gets a unique identity tied to its owner's domain.
Delegation
Developer Tools
To Full Agent Trust
Create an agent, delegate permissions, and verify trust — the entire flow in 5 lines of TypeScript.
Gzipped Bundle
Tree-shakeable ESM architecture. Import only what you need — identity, delegation, or verification.
Open Source
Apache-2.0 licensed. No vendor lock-in, no hidden fees. The only open-source, developer-first agent identity SDK.
Built for Agent Developers
“We needed a way to prove our agents were authorized. Credat gave us that in an afternoon. Dead simple API.”
“The delegation model is exactly what we needed. Scoped permissions for agents that just work. No overengineering.”
“We run a multi-agent platform. Credat solved our biggest problem: knowing which agent is actually authorized to do what.”
“TypeScript-first, open source, and it handles the hard part — trust between agents — so we don't have to.”
“We evaluated building our own agent auth system. Credat saved us months and is more secure than what we would have built.”
“Finally someone is solving agent identity properly. Open source, developer-first, and built on real standards under the hood.”
Stop Agent Fraud. Start Shipping Trust.
Credat is the open-source trust layer for AI agents. Give your agents an identity, delegate permissions, and verify trust — in 5 lines of code.