AgentDID gives every AI agent a permanent, verifiable on-chain identity — with an immutable record of every action it takes.
Non-commercial · Open source · Community governed · No tokens · No investors
Agents Registered
Actions On-Chain
Anomalies Flagged
Latest Block
Each agent gets a unique Ethereum address as its DID — verifiable by anyone, owned by no single authority.
Every query, tool use, and decision is hashed and logged on-chain. The past cannot be rewritten.
Anyone can audit any agent's history or flag anomalies. Accountability without a central authority.
From passport-style cross-platform identity to autonomous subscriptions, transactions, and compliance audits — AgentDID is the identity layer every AI agent ecosystem needs.
Agent Passport
Every AI agent gets a permanent DID (did:agent:<address>) that travels with it across platforms, frameworks, and organizations — like a passport for the digital world.
// Any framework, same identity
const passport = await AgentPassport.create(rpcUrl, privateKey)
// → did:agent:0xABCD...
// Works on LangChain, AutoGPT, CrewAI, custom agents
await passport.logAction("tool_use", inputHash, outputHash, true)All use cases work today on Ethereum Sepolia — register your agent above to get started.
Any language, any AI framework. Your agent gets a permanent on-chain identity with two function calls.
import { AgentPassport } from "@agentdid/sdk"
const passport = await AgentPassport.create({
registryAddress: "0x05623871958D6d648953e64B1cdb562Adc28019B",
rpcUrl: "https://rpc.sepolia.org",
}, process.env.AGENT_PRIVATE_KEY)
// Build capability declaration
const cap = AgentPassport.buildCapabilityDeclaration({
name: "MyBot",
version: "1.0.0",
purpose: "Research assistant — web search and summarization",
allowedActionTypes: ["llm_query", "web_search"],
riskLevel: RiskLevel.LOW,
category: AgentCategory.RESEARCH,
})
// Register on-chain (one time)
await passport.register("MyBot", "1.0.0", cap)
// → did:agent:0xABCD...
// Log every action automatically
await passport.logAction("llm_query", userPrompt, response, true)All registered agents on the AgentDID protocol — publicly auditable by anyone.