About x402-factory
x402-factory provides typed clients, examples, and an agent skill for paid data APIs covering markets, weather, economics, SEC filings, and recall feeds. It also publishes free verification and calibration data; API calls use x402 payments.
Key features
- 01Typed TypeScript and Python clients
- 02Free forecast verification scores
- 03Public market calibration data
- 04Agent skill with free and paid modes
- 05Per-call x402 payment support
Use cases
- 01Inspect forecast performance before use
- 02Query paid data APIs from an agent
- 03Discover available feeds without a wallet
MCP support
Provides a remote MCP endpoint over stateless streamable HTTP. Free tools return data, while paid tools return a structured payment-required pointer with the HTTP route and price.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://x402-factory.com/mcpclaude mcp add --transport http praneeth12098-x402-factory https://x402-factory.com/mcp{"mcpServers":{"praneeth12098-x402-factory":{"type":"http","url":"https://x402-factory.com/mcp"}}}Open in Cursor{"servers":{"praneeth12098-x402-factory":{"type":"http","url":"https://x402-factory.com/mcp"}}}Open in VS Code{"mcpServers":{"praneeth12098-x402-factory":{"type":"http","url":"https://x402-factory.com/mcp"}}}{"mcpServers":{"praneeth12098-x402-factory":{"serverUrl":"https://x402-factory.com/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEx402-factory clients
Typed clients, worked examples, and an agent skill for x402-factory.com — 200+ paid data APIs for AI agents (prediction-market probabilities, weather intelligence, US economic prints, SEC/EDGAR and recall feeds), sold per-call via the x402 protocol (HTTP 402 + USDC on Base). No API keys, no accounts: cents per call from any wallet.
Judge it before you pay — the free proof surfaces
The platform publishes its own scorecards, free forever:
- Forecast verification — every forecast scored against real NWS
station observations, segmented by lead/region/method/model-version:
/v1/verification/scores. Under the current model the station blend beats raw NWS at short leads (as of 2026-10-15, 90-day rolling window, on thenws-blendsegment — the same-sample paired comparison against the NWS baseline: temperature MAE 0.84 vs 1.14 °C at 1-hour lead, n=3,506; 0.89 vs 1.16 at 3h; converging to NWS parity by 24h — the decay the method predicts. Recompute it yourself: filter the live table to segmentnws-blend). - Model-vs-market calibration gate — market classes are sold only
when their own resolved history proves calibrated; the full reliability
table (and what is currently withheld, with reasons) is public:
/v1/edge/calibration. - Free delayed feeds and change-probes on the probability catalog.
Quickstart (TypeScript)
import { X402FactoryClient } from "./ts/client.js";
// Free routes need nothing:
const free = new X402FactoryClient();
console.log(await free.verificationScores());
// Paid routes need any funded wallet via @x402/fetch:
import { wrapFetchWithPayment } from "@x402/fetch";
import { privateKeyToAccount } from "viem/accounts";
const paid = new X402FactoryClient({
fetch: wrapFetchWithPayment(fetch, privateKeyToAccount(process.env.PK as `0x${string}`)),
});
console.log(await paid.get("/v1/econ/dashboard")); // $0.03, settles automatically
Python: see py/client.py. Worked examples (free routes
only, run as-is): examples/.
MCP
Remote MCP endpoint: POST https://x402-factory.com/mcp (streamable HTTP,
stateless JSON). Free and delayed tools return data; paid tools return a
structured payment_required pointer with the HTTP route and price — the
mirror is a free funnel, not a paywall bypass. Rate limit: 30 req/min.
Agent skill
skill/ packages the catalog as a task-oriented skill:
discovery mode with no wallet (free routes only) and paid mode through
your existing x402 wallet tooling with a per-call cap. MIT-0.
Honesty contract
Every response carries meta.sources and meta.data_mode; method labels
are end-to-end truthful (NWS passthrough is labeled as such); skill
claims appear only where measured, with n, on the free proof surfaces.
License: MIT.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.