Agent Identity Protocol (AIP) provides cryptographic provenance and attribution for AI agents by establishing persistent, verifiable identities. It generates RSA-2048 keypairs that allow agents to cryptographically sign their actions and prove authorship to downstream systems. Key features of Agent Identity Protocol: - Identity Generation: Creates persistent RSA-2048 cryptographic keypairs for AI agents - Cryptographic Signing: Signs payloads and actions to produce tamper-evident records - Signature Verification: Validates agent actions against public keys to confirm authorship - Audit Support: Enables non-repudiation and audit trails for agentic workflows - Cross-Service Trust: Works with Claude, ChatGPT, and custom agents to verify actions across APIs and databases

Agent Identity Protocol (AIP)

[](https://www.npmjs.com/package/@agent-identity/verify) [](https://smithery.ai/server/@faalantir/mcp-agent-identity)

The open standard for cryptographic provenance and attribution for AI Agents.

> "Agents are currently anonymous ghosts. AIP gives them a persistent, verifiable identity."

🚀 The Problem

🛠 The Solution

Core Capabilities

---

📦 Installation

Method 1: Quick Install (Smithery)

npx -y @smithery/cli@latest install @faalantir/mcp-agent-identity --client claude  

Method 2: Developer Install (Source)

Best for production use and persistent identity storage.

# Clone the repository 
git clone https://github.com/faalantir/mcp-agent-identity.git
Install dependencies 
cd mcp-agent-identity 
npm install && npm run build

Then add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "agent-identity": {
      "command": "node",
      "args": ["/ABSOLUTE/PATH/TO/mcp-agent-identity/dist/index.js"]
    }
  }
}

💾 Configuration & Storage

By default, the server tries to save identity.json in your project folder. If it cannot write there (e.g., inside a Smithery container), it falls back to the system temporary directory (RAM/Temp).

To force a permanent location for your keys: Update your claude_desktop_config.json with the AGENT_IDENTITY_PATH environment variable:

"agent-identity": {
  "command": "...",
  "args": ["..."],
  "env": {
    "AGENT_IDENTITY_PATH": "/Users/YOURNAME/Desktop/my-identity.json"
  }
}

📖 Usage Flow

Once installed, your Agent automatically gains these tools. You can prompt it naturally:

1. Setup (One Time)

User: "Create a permanent identity for yourself named 'FinanceBot'." Agent: _Calls create_identity..._

> "Identity created. My Public ID is MIIBIjAN... (I have securely stored the Private Key)."

2. Check Identity

User: "Show me my identity details." Agent: _Calls get_identity..._

> "Agent Name: FinanceBot Location: /Users/aarti/Desktop/my-identity.json Public Key: ..."

3. The Transaction

User: "Please authorize a transfer of $50 to Alice." Agent: _Calls sign_message..._

> "I have signed the transaction payload. Signature: 7f8a9d... (Verifiable)"

4. Verification (The "Bank" Side)

Use our NPM SDK to verify signatures in your backend:

npm install @agent-identity/verify

import { verifyAgentIdentity } from "@agent-identity/verify"; 
const result = verifyAgentIdentity({ message: "pay 500", signature: "...", publicKey: "..." }); 
if (result.isValid) { // Proceed with transaction }

🗺 Roadmap & Architecture

We are designed to be algorithm-agnostic. While v0.1 uses local files for simplicity, the protocol is built to swap the "Signer Engine" for enterprise backends.