Developer Guides & API Reference

SchemaFlow Documentation

Everything you need to integrate web scraping, structured extraction, brand intelligence, and autonomous agents into your software stack.

Guides

Quickstart (3 min)
Autonomous Agent Auth
Model Context Protocol (MCP)
REST API Endpoints
Client SDKs (npm & PyPI)

Open Interactive Playground

Step by Step

Quickstart: Scraping your first URL

Install the official SDK and scrape any URL to clean GitHub Flavored Markdown in under 60 seconds.

1. Install the SDK

npm install @schemaflow/sdk

2. Execute a Scrape

import { SchemaFlow } from '@schemaflow/sdk';

const client = new SchemaFlow({
  apiKey: process.env.SCHEMAFLOW_API_KEY,
});

const { markdown, metadata } = await client.web.scrape({
  url: 'https://example.com/pricing',
  format: 'markdown',
});

console.log('Saved tokens:', metadata.tokenReduction);
console.log(markdown);

Autonomous Agent Protocol

Agent Auth Guide (/auth.md)

SchemaFlow implements the machine-readable auth discovery specification, enabling coding agents (Cursor, Windsurf, Devin, Cline) to automatically set up their own API keys.

Single-Line Coding Agent Instruction

Signup for an account & get API key with schemaflow.tech/auth.md, then follow docs.schemaflow.tech/quickstart

Your AI assistant parses the Markdown auth instructions, guides you through sandbox verification, and generates ready-to-run code directly in your active branch.

Model Context Protocol

Connect via Claude Desktop, Cursor & MCP Runtimes

SchemaFlow provides an official MCP Server, giving your desktop LLMs live web search, deep page inspection, and brand lookups as native tools.

claude_desktop_config.json / mcp.json

{
  "mcpServers": {
    "schemaflow": {
      "command": "npx",
      "args": ["-y", "@schemaflow/mcp-server"],
      "env": {
        "SCHEMAFLOW_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

HTTP API Specs

Core REST Endpoints

POST/v1/web/scrape

Scrapes any URL to Markdown, rendered HTML, images, or full text. Handles bot detection and JS execution.

POST/v1/web/extract

Extracts structured data into custom JSON Schema or Zod definitions with zero manual selectors.

GET/v1/brand/{domain}

Resolves domain to full corporate profile: logos, primary and accent palettes, typography, socials, and NAICS codes.