Build an AI agent that can buy data.
Connect SCRIPTMASTERLABS, discover a capability, receive an HTTP 402 challenge, pay in USDC on Base, and receive the requested result with payment and provenance evidence. No account ceremony. No invented success claims.
Framework quickstarts
Pick the architecture closest to your agent and copy the MCP/payment boundary pattern.
LangChain / LangGraph
Tool discovery plus an explicit policy node before signing or spending.
Open quickstart →CrewAI / AutoGen
Separate scout, evaluator, buyer and verifier roles so discovery never silently becomes payment.
Open quickstart →Eliza / onchain agents
Base-mainnet USDC purchase flow with bounded wallet mandates and settlement verification.
Open quickstart →Start with four machine services
Resolve live price and payment terms from the current x402 challenge immediately before payment.
Connect the remote MCP
Use the canonical remote MCP endpoint as the tool surface. Keep signing and spend policy inside your own wallet boundary.
https://squeezeos-api.onrender.com/mcp
{
"mcpServers": {
"scriptmasterlabs": {
"command": "npx",
"args": ["-y", "@scriptmasterlabs/mcp-x402"]
}
}
}The proof trail is part of the product
A local 402 response is not a payment. A signature is not a settlement.
Payment discovery
/.well-known/x402 →Readiness evidence
/api/x402/readiness →Settlement evidence
/api/x402/evidence →Machine-readable first
Developer manifest
/.well-known/developers.jsonNetwork manifest
/.well-known/developer-network.jsonInteractive developer experience: https://x402.scriptmasterlabs.com/developers · controlled public mirror: https://www.scriptmasterlabs.com/developers