PROTOCOL V1.0 LIVE
Don't trust the agent.
Trust the human behind it.
AgentCred is an open-source protocol that cryptographically proves which human is responsible for an AI agent's actions.
The Problem: Anonymous Agents
AI agents are everywhere, but they lack identity. When an agent posts, comments, or acts, you have no way to verify who is controlling it. This leads to manipulation, spam, and deep distrust.
- ✕ Reddit can't tell who's a bot.
- ✕ YouTube can't tell who's commenting.
- ✕ You can't tell who to trust.
🤖
UNVERIFIED SOURCE
🤖
VERIFIED: @GITHUB_USER
The Solution: Accountability
Like X's blue badge, but for AI agents. AgentCred provides a standard protocol for agents to sign their actions, linking them undeniably to a human identity via GitHub.
- ✓ Cryptographic proof (Ed25519 signatures)
- ✓ Backed by GitHub identity
- ✓ Completely Open Source (MIT)
How It Works
01
Install
Add the package to your agent's codebase.
npm install agentcred 02
Sign
Your agent signs its output using your credentials.
agent.sign(output) 03
Verify
Anyone can verify the signature and see the human behind it.
verify(envelope) Quick Start
example.ts
import { createIdentity, sign, verify } from 'agentcred'
// 1. Create identity (linked to GitHub)
const identity = await createIdentity(githubToken)
// 2. Sign agent output
const envelope = await sign("Hello from my AI agent!", identity)
// 3. Verify anywhere
const result = await verify(envelope)
console.log(result.verified) // true
console.log(result.github.username) // "@yourname"See the badge in your browser
Coming soon to the Chrome Web Store. Verify agents wherever they post.
https://social-media.com/post/123
Verified
Signed by
@github_user