Trusty Squire MCP server
{
"mcpServers": {
"trusty-squire": {
"command": "npx",
"args": [
"-y",
"@trusty-squire/mcp@latest"
]
}
}
}{
"mcpServers": {
"trusty-squire": {
"command": "npx",
"args": [
"-y",
"@trusty-squire/mcp@latest"
]
}
}
}{
"mcpServers": {
"trusty-squire": {
"command": "npx",
"args": [
"-y",
"@trusty-squire/mcp@latest"
]
}
}
}{
"servers": {
"trusty-squire": {
"command": "npx",
"args": [
"-y",
"@trusty-squire/mcp@latest"
]
}
}
}{
"mcpServers": {
"trusty-squire": {
"command": "npx",
"args": [
"-y",
"@trusty-squire/mcp@latest"
]
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsoninstall npx -y @trusty-squire/mcpversion 1.1.16-rc.6
What it does
Trusty Squire is an MCP server that automates account creation, sign-in, setup flows, and payments on websites by operating a real browser. It is designed for coding agents and developers who need to set up OAuth, API keys, manage payments, and handle other web-based provisioning tasks without exposing credentials or payment cards in agent context.
- Set up OAuth clients (like Google OAuth) for an application and store the generated secret securely without exposing it in chat or code.
- Complete payment checkouts and purchases on behalf of a user while keeping payment card details encrypted and requiring approval via passkey.
- Create and manage API keys for third-party services (like Stripe or Render) and issue scoped backend grants without storing plaintext secrets in agent context.
Tools this server exposes
| Tool | Does |
|---|---|
| operate_pay | Complete a checkout on a website by reading the total amount and currency, sending an approval link to the user, and submitting the payment after approval. |
| fetch_credential | Retrieve a stored plaintext credential after the user signs the request with their passkey. |
| reveal | Show the masked number, name, and expiry of a saved payment card after running a passkey ceremony in the user's browser. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.