Frisk MCP server
{
"mcpServers": {
"tryfrisk-frisk": {
"command": "npx",
"args": [
"-y",
"frisk-mcp@latest"
]
}
}
}{
"mcpServers": {
"tryfrisk-frisk": {
"command": "npx",
"args": [
"-y",
"frisk-mcp@latest"
]
}
}
}{
"mcpServers": {
"tryfrisk-frisk": {
"command": "npx",
"args": [
"-y",
"frisk-mcp@latest"
]
}
}
}{
"servers": {
"tryfrisk-frisk": {
"command": "npx",
"args": [
"-y",
"frisk-mcp@latest"
]
}
}
}{
"mcpServers": {
"tryfrisk-frisk": {
"command": "npx",
"args": [
"-y",
"frisk-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 frisk-mcpversion 0.0.1
What it does
Frisk screens counterparties before x402 payments or tool calls, returning a verdict (allow, review, or block) with a trust score and reasons. It runs offline by default in lite mode with no dependencies, and optionally connects to a hosted API for reputation history and threat intelligence.
- Check whether a seller address is trustworthy before an agent makes a payment to them.
- Verify that a payment endpoint is served over HTTPS and the requested payTo matches the intended counterparty.
- Enforce payment policies by screening transaction amounts and assets against configurable limits before execution.
Tools this server exposes
| Tool | Does |
|---|---|
| screen_payment | Screens a counterparty address and transaction details, returning a verdict (allow, review, or block) with trust score and reasons. |
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.