Skip to main content
MCP server CX Codex

Agents Trust — Claude Code MCP server: what it does, how to install

Agents Trust discovers APIs tailored to task prompts and inspects verified endpoint evidence using Base USDC micropayments.

MIT updated 2026-10-27

Install into Claude Code · Platform CLI All methods and platforms ↓
claude mcp add --env AGENTS_TRUST_ENABLE_PAYMENTS=<AGENTS_TRUST_ENABLE_PAYMENTS> --env AGENTS_TRUST_WALLET_PRIVATE_KEY=<AGENTS_TRUST_WALLET_PRIVATE_KEY> --env AGENTS_TRUST_MAX_CALL_USDC=<AGENTS_TRUST_MAX_CALL_USDC> --env AGENTS_TRUST_SESSION_CAP_USDC=<AGENTS_TRUST_SESSION_CAP_USDC> --transport stdio agents-trust -- npx -y @agents-trust/mcp
Open on GitHub

What Agents Trust does

Agents Trust matches engineering prompts to functional APIs directly inside the MCP host. The recommend tool returns target URLs, pricing figures, and call formats. When you need to verify an endpoint before integrating, inspect retrieves cached test records without contacting the external provider.

The stdio process runs locally on Node.js 20 or higher via npx. Diagnostics via health, payment confirmations through receipt, and demo recommendations work without financial settings. Paid lookups use USDC on Base mainnet (eip155:8453) through local x402 signing delivered only to https://api.agents-trust.com.

Transactions stay blocked until you supply AGENTS_TRUST_ENABLE_PAYMENTS=true. Every runtime session allows only one payment authorization under the AGENTS_TRUST_SESSION_CAP_USDC cap. Networks other than Base mainnet will fail.

What it helps with

  • find third-party APIs matching functional task descriptions
  • inspect historical reliability records for a recommended endpoint
  • settle discovery micro-fees with a local Base USDC wallet
  • track payment confirmations using transaction attempt identifiers

Example prompts for the agent

  • Find weather forecast endpoints with compact output using the recommend tool
  • Inspect the stored reliability evidence for a recommended API URL
  • Check settlement status for payment_attempt_id using receipt

How to install Agents Trust

Platform CLI

claude mcp add --env AGENTS_TRUST_ENABLE_PAYMENTS=<AGENTS_TRUST_ENABLE_PAYMENTS> --env AGENTS_TRUST_WALLET_PRIVATE_KEY=<AGENTS_TRUST_WALLET_PRIVATE_KEY> --env AGENTS_TRUST_MAX_CALL_USDC=<AGENTS_TRUST_MAX_CALL_USDC> --env AGENTS_TRUST_SESSION_CAP_USDC=<AGENTS_TRUST_SESSION_CAP_USDC> --transport stdio agents-trust -- npx -y @agents-trust/mcp

Replace <…> with your own values

Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more

Questions about Agents Trust

How to install Agents Trust in Claude Code?

Run `claude mcp add --env AGENTS_TRUST_ENABLE_PAYMENTS=<AGENTS_TRUST_ENABLE_PAYMENTS> --env AGENTS_TRUST_WALLET_PRIVATE_KEY=<AGENTS_TRUST_WALLET_PRIVATE_KEY> --env AGENTS_TRUST_MAX_CALL_USDC=<AGENTS_TRUST_MAX_CALL_USDC> --env AGENTS_TRUST_SESSION_CAP_USDC=<AGENTS_TRUST_SESSION_CAP_USDC> --transport stdio agents-trust -- npx -y @agents-trust/mcp`.

What does Agents Trust do?

It queries catalog APIs based on task descriptions and displays stored performance evidence. Paid queries draw USDC directly from your own Base mainnet wallet.

Are wallet keys mandatory for evaluation?

No. The health and receipt tools run free of charge, and free demo recommendations work without wallet keys.

Similar MCP servers

All MCP servers for Agent setup: memory, context, orchestration →

Project sponsors

Companies that keep this analytics open