Grabs Vercel Model Capabilities and creates a modelData array for mastra.ai docs.

index.js

import * as cheerio from "cheerio";

const apiKeyMap = {
  "xAI Grok": "XAI_API_KEY",
  OpenAI: "OPENAI_API_KEY",
  Anthropic: "ANTHROPIC_API_KEY",
  Mistral: "MISTRAL_API_KEY",
  "Google Generative AI": "GOOGLE_API_KEY",
  "Google Vertex": "GOOGLE_VERTEX_API_KEY",
  DeepSeek: "DEEPSEEK_API_KEY",
  Cerebras: "CEREBRAS_API_KEY",
  Groq: "GROQ_API_KEY",
  Vercel: "VERCEL_API_KEY"
};

function parseBool(cell) {
  const val = cell.toLowerCase();
  if (val.includes("<check")) return true;
  if (val.includes("<cross")) return false;
  return val === "yes";
}

// Helper: Get first word lowercased for provider URL slug
function slugifyProviderFirstWord(name) {
  const firstWord = name.trim().split(/\s+/)[0];
  return firstWord.toLowerCase();
}

function getApiKey(provider) {
  // Extract plain text from markdown link [Name](url)
  const plainProvider = provider.replace(/\[([^\]]+)\]\([^)]+\)/, "$1").trim();

  // Find matching key ignoring case
  const key = Object.keys(apiKeyMap).find((k) => k.toLowerCase() === plainProvider.toLowerCase());

  return key ? apiKeyMap[key] : "";
}

(async () => {
  const url = "https://raw.githubusercontent.com/vercel/ai/refs/heads/main/content/docs/02-foundations/02-providers-and-models.mdx";
  const res = await fetch(url);
  const mdx = await res.text();

  const sectionRegex = /## Model Capabilities\s+([\s\S]*?)(?=\n## |\n$)/;
  const match = sectionRegex.exec(mdx);

  if (!match) {
    console.error("❌ Could not find the Model Capabilities section");
    return;
  }

  const tableMarkdown = match[1].trim();

  const tableMatch = tableMarkdown.match(/\|(.+\|)+\n\|[-|:\s]+\|\n([\s\S]+?)(?=\n\n|\n$)/);
  if (!tableMatch) {
    console.error("❌ Could not find a markdown table in the section");
    return;
  }

  const headerRow = tableMatch[0].split("\n")[0];
  const dataRows = tableMatch[0].split("\n").slice(2);

  const htmlTable = `
    <table>
      <thead>
        <tr>${headerRow
          .split("|")
          .filter(Boolean)
          .map((c) => `<th>${c.trim()}</th>`)
          .join("")}</tr>
      </thead>
      <tbody>
        ${dataRows
          .map((row) => {
            const cells = row.split("|").filter(Boolean);
            return `<tr>${cells.map((c) => `<td>${c.trim()}</td>`).join("")}</tr>`;
          })
          .join("")}
      </tbody>
    </table>
  `;

  const $ = cheerio.load(htmlTable);
  const modelData = [];

  $("tbody tr").each(function () {
    const tds = $(this).find("td");
    const rawProvider = tds.eq(0).html() || "";
    const provider = rawProvider.replace(/\[([^\]]+)\]\([^)]+\)/, "$1").trim();

    let model = tds.eq(1).text().trim();
    model = model.replace(/`/g, "");

    const imageInput = parseBool(tds.eq(2).html() || "");
    const objectGeneration = parseBool(tds.eq(3).html() || "");
    const toolUsage = parseBool(tds.eq(4).html() || "");
    const toolStreaming = parseBool(tds.eq(5).html() || "");

    const apiKey = getApiKey(provider);
    const providerSlug = slugifyProviderFirstWord(provider);

    modelData.push({
      provider,
      providerUrl: `https://sdk.vercel.ai/providers/ai-sdk-providers/${providerSlug}`,
      model,
      imageInput,
      objectGeneration,
      toolUsage,
      toolStreaming,
      apiKey
    });
  });

  console.log("const modelData =", JSON.stringify(modelData, null, 2));
})();