Skip to main content

​
MCP Services

Cred Protocol provides Model Context Protocol (MCP) services that enable AI agents and assistants to access on-chain credit scoring, financial reporting, identity attestation, and agentic reputation data.
MCP is an open protocol that allows AI systems to securely interact with external tools and data sources. Learn more about MCP →

​
What is MCP?

The Model Context Protocol (MCP) is a standardized way for AI agents (like Claude, GPT, or custom LLM applications) to:
  • Discover available tools and their capabilities
  • Call tools with structured parameters
  • Receive structured responses
Unlike traditional REST APIs that require manual integration, MCP allows AI agents to automatically understand and use your tools through natural language.

​
Why Use MCP Services?

AI-Native Integration

AI agents can discover and use Cred Protocol tools without custom code

Natural Language

Users can ask “What’s this wallet’s credit score?” and the agent handles it

Sandbox Data

Safe testing with deterministic mock data - no real blockchain calls

No Authentication

MCP endpoints don’t require API tokens for easy development

​
Available Tools

​
Sandbox Tools (Mock Data)

ToolDescription
get_credit_scoreGet credit score (300-1000) for an Ethereum address
get_credit_scores_batchGet scores for multiple addresses at once
get_financial_summaryComprehensive financial report with assets, DeFi positions, and activity
get_identity_attestationsVerified identity credentials (ENS, Gitcoin Passport, POAPs)
get_portfolio_valueTotal portfolio value in USD across all chains
get_chain_portfolio_valuePortfolio value on a specific blockchain

​
Live Tools (Real Data)

ToolDescription
get_live_credit_scoreReal credit score using the Andromeda 1.0 model with on-chain data (requires API key)
get_live_financial_summaryReal multi-chain financial report with assets, DeFi, and percentiles (requires API key)
get_live_identity_attestationsReal identity attestations from blockchain (requires API key)
search_agentsSearch AI agents in the ERC-8004 Identity Registry on Base (requires API key)
Live tools require the CRED_API_KEY environment variable. Generate an API key from the Cred Protocol Dashboard.

​
Agentic Reputation (ERC-8004)

In addition to wallet credit scoring, Cred Protocol MCP services provide access to the ERC-8004 Identity Registry — an on-chain registry of AI agents deployed on Base. The search_agents tool allows AI agents and applications to discover other registered agents by name or description, returning agent IDs, owner addresses, and registration metadata. This is useful for:
  • Agent discovery — Find agents by capability or domain
  • Trust verification — Look up an agent’s on-chain registration before interacting with it
  • Ecosystem mapping — Explore the landscape of registered AI agents
ERC-8004 defines a standard for registering AI agent identities on-chain as ERC-721 tokens. Each agent has an owner address and metadata (name, description, image) stored via tokenURI. The registry lives on Base (chain ID 8453).
Unlike credit scoring tools which assess wallet reputation, the agent search tool operates on the agent layer — helping AI systems reason about other AI agents in the ecosystem.

​
Access Methods

Cred Protocol MCP services can be accessed two ways:

​
1. HTTP/REST Endpoints

Standard HTTP endpoints at /mcp/sandbox/* - works with any HTTP client: 
curl https://api.credprotocol.com/mcp/sandbox/score/vitalik.eth

​
2. MCP Protocol (stdio)

Native MCP protocol for AI agents like Cursor, Claude Desktop, or custom MCP clients: 
{
  "mcpServers": {
    "cred-protocol": {
      "command": "python",
      "args": ["mcp_server.py"]
    }
  }
}

​
Agent Skill Instructions

When an AI agent connects via MCP, the server automatically provides skill instructions that guide the agent on:
  • Which tool to use for different user intents
  • How to interpret credit scores, financial summaries, and attestations
  • When to escalate from a quick score check to a full financial report
  • Best practices like batching requests and flagging red flags
These instructions are delivered to agents during MCP initialization — no extra configuration needed. See the full Agent Skill reference.

​
MCP vs REST API

FeatureMCP ServicesREST API
DataSandbox/mockReal blockchain data
AuthenticationNone requiredAPI token required
PurposeDevelopment, AI agentsProduction applications
BillingFreeConsumes Cred Units
Response TimeInstant (~10ms)3-10 seconds
MCP services use sandbox data for safe testing. For production applications with real blockchain data, use the REST API.

​
Next Steps

Getting Started

Connect your AI agent to Cred Protocol MCP

Available Tools

Explore all available MCP tools

Examples

See practical usage examples

REST API

Use the production API for real data