nodebench-mcp 2.17.0 → 2.18.0

package/dist/tools/rssTools.js

@@ -8,25 +8,25 @@ import { getDb } from "../db.js";
 // ── SQLite schema ────────────────────────────────────────────────────────────
 function ensureRssTables() {
     const db = getDb();
-    db.exec(`
-    CREATE TABLE IF NOT EXISTS rss_sources (
-      id INTEGER PRIMARY KEY AUTOINCREMENT,
-      url TEXT NOT NULL UNIQUE,
-      name TEXT,
-      category TEXT,
-      created_at TEXT DEFAULT (datetime('now'))
-    );
-    CREATE TABLE IF NOT EXISTS rss_articles (
-      id INTEGER PRIMARY KEY AUTOINCREMENT,
-      source_url TEXT NOT NULL,
-      title TEXT,
-      link TEXT NOT NULL,
-      published TEXT,
-      summary TEXT,
-      fetched_at TEXT DEFAULT (datetime('now')),
-      is_new INTEGER DEFAULT 1,
-      UNIQUE(source_url, link)
-    );
+    db.exec(`
+    CREATE TABLE IF NOT EXISTS rss_sources (
+      id INTEGER PRIMARY KEY AUTOINCREMENT,
+      url TEXT NOT NULL UNIQUE,
+      name TEXT,
+      category TEXT,
+      created_at TEXT DEFAULT (datetime('now'))
+    );
+    CREATE TABLE IF NOT EXISTS rss_articles (
+      id INTEGER PRIMARY KEY AUTOINCREMENT,
+      source_url TEXT NOT NULL,
+      title TEXT,
+      link TEXT NOT NULL,
+      published TEXT,
+      summary TEXT,
+      fetched_at TEXT DEFAULT (datetime('now')),
+      is_new INTEGER DEFAULT 1,
+      UNIQUE(source_url, link)
+    );
   `);
 }
 /** Extract text content from an XML tag, handling CDATA */
@@ -266,14 +266,14 @@ export const rssTools = [
             const format = args.format || "markdown";
             // Query new articles
             const params = [sinceHours];
-            let query = `
-        SELECT a.title, a.link, a.published, a.summary, a.source_url, a.fetched_at,
-               COALESCE(s.name, a.source_url) as source_name,
-               COALESCE(s.category, 'uncategorized') as category
-        FROM rss_articles a
-        LEFT JOIN rss_sources s ON a.source_url = s.url
-        WHERE a.is_new = 1
-        AND a.fetched_at >= datetime('now', '-' || ? || ' hours')
+            let query = `
+        SELECT a.title, a.link, a.published, a.summary, a.source_url, a.fetched_at,
+               COALESCE(s.name, a.source_url) as source_name,
+               COALESCE(s.category, 'uncategorized') as category
+        FROM rss_articles a
+        LEFT JOIN rss_sources s ON a.source_url = s.url
+        WHERE a.is_new = 1
+        AND a.fetched_at >= datetime('now', '-' || ? || ' hours')
       `;
             if (category) {
                 query += " AND s.category = ?";
@@ -283,9 +283,9 @@ export const rssTools = [
             const articles = db.prepare(query).all(...params);
             // Mark articles as seen
             const markParams = [sinceHours];
-            let markQuery = `
-        UPDATE rss_articles SET is_new = 0
-        WHERE is_new = 1 AND fetched_at >= datetime('now', '-' || ? || ' hours')
+            let markQuery = `
+        UPDATE rss_articles SET is_new = 0
+        WHERE is_new = 1 AND fetched_at >= datetime('now', '-' || ? || ' hours')
       `;
             if (category) {
                 markQuery += ` AND source_url IN (SELECT url FROM rss_sources WHERE category = ?)`;
@@ -326,24 +326,24 @@ export const rssTools = [
             }
             if (format === "html") {
                 const sections = [...byCategory.entries()]
-                    .map(([cat, items]) => `
-          <h2 style="color:#333;border-bottom:2px solid #007bff;padding-bottom:4px">${cat} (${items.length})</h2>
+                    .map(([cat, items]) => `
+          <h2 style="color:#333;border-bottom:2px solid #007bff;padding-bottom:4px">${cat} (${items.length})</h2>
           ${items
-                    .map((a) => `
-            <div style="margin-bottom:16px;padding:12px;background:#f8f9fa;border-radius:6px">
-              <a href="${a.link}" style="font-size:16px;font-weight:600;color:#007bff;text-decoration:none">${a.title}</a>
-              <div style="color:#666;font-size:12px;margin-top:4px">${a.source_name} · ${a.published || a.fetched_at}</div>
-              ${a.summary ? `<p style="margin-top:8px;color:#444;font-size:14px">${a.summary}</p>` : ""}
+                    .map((a) => `
+            <div style="margin-bottom:16px;padding:12px;background:#f8f9fa;border-radius:6px">
+              <a href="${a.link}" style="font-size:16px;font-weight:600;color:#007bff;text-decoration:none">${a.title}</a>
+              <div style="color:#666;font-size:12px;margin-top:4px">${a.source_name} · ${a.published || a.fetched_at}</div>
+              ${a.summary ? `<p style="margin-top:8px;color:#444;font-size:14px">${a.summary}</p>` : ""}
             </div>`)
                     .join("")}`)
                     .join("");
-                const html = `
-          <div style="font-family:-apple-system,BlinkMacSystemFont,sans-serif;max-width:640px;margin:0 auto;padding:20px">
-            <h1 style="color:#111;margin-bottom:4px">Research Digest</h1>
-            <p style="color:#666;margin-top:0">${articles.length} new articles · ${new Date().toLocaleDateString()}</p>
-            ${sections}
-            <hr style="margin-top:32px;border:none;border-top:1px solid #ddd">
-            <p style="color:#999;font-size:11px">Generated by NodeBench MCP · build_research_digest</p>
+                const html = `
+          <div style="font-family:-apple-system,BlinkMacSystemFont,sans-serif;max-width:640px;margin:0 auto;padding:20px">
+            <h1 style="color:#111;margin-bottom:4px">Research Digest</h1>
+            <p style="color:#666;margin-top:0">${articles.length} new articles · ${new Date().toLocaleDateString()}</p>
+            ${sections}
+            <hr style="margin-top:32px;border:none;border-top:1px solid #ddd">
+            <p style="color:#999;font-size:11px">Generated by NodeBench MCP · build_research_digest</p>
           </div>`;
                 return [
                     {
@@ -427,299 +427,299 @@ export const rssTools = [
                 ? feeds.map((f) => `  { url: "${f.url}", name: "${f.name || f.url}", category: "${f.category || "general"}" },`).join("\n")
                 : `  // Add your feeds here:\n  // { url: "https://arxiv.org/rss/cs.AI", name: "arXiv AI", category: "ai-research" },\n  // { url: "https://hnrss.org/newest?points=100", name: "Hacker News Top", category: "tech" },`;
             // ── Generate the main script ──
-            const mainScript = `#!/usr/bin/env node
-/**
- * ${projectName} — Automated Research Digest Pipeline
- *
- * Subscribe to RSS/Atom feeds, fetch new articles, build a digest, and email it.
- * Generated by NodeBench MCP scaffold_research_pipeline.
- *
- * Usage:
- *   node digest.mjs                    # Run once (fetch + digest + email)
- *   node digest.mjs --fetch-only       # Just fetch, don't email
- *   node digest.mjs --list-sources     # Show registered feeds
- *   node digest.mjs --add-feed <url>   # Add a new feed
- *
- * Schedule with cron:
- *   crontab -e
- *   ${cronExpr[schedule]} cd /path/to/${projectName} && node digest.mjs >> digest.log 2>&1
- */
-
-import { readFileSync, writeFileSync, existsSync, mkdirSync } from "node:fs";
-import * as tls from "node:tls";
-
-// ── Config ────────────────────────────────────────────────────────────────────
-
-const DATA_DIR = new URL("./data/", import.meta.url).pathname.replace(/^\\//, "");
-if (!existsSync(DATA_DIR)) mkdirSync(DATA_DIR, { recursive: true });
-
-const SOURCES_FILE = DATA_DIR + "sources.json";
-const ARTICLES_FILE = DATA_DIR + "articles.json";
-
-const EMAIL_USER = process.env.EMAIL_USER || "";
-const EMAIL_PASS = process.env.EMAIL_PASS || "";
-const EMAIL_TO = process.env.DIGEST_TO || "${emailTo}" || EMAIL_USER;
-const SMTP_HOST = process.env.EMAIL_SMTP_HOST || "smtp.gmail.com";
-const SMTP_PORT = parseInt(process.env.EMAIL_SMTP_PORT || "465");
-
-// ── Feed storage (JSON file-based, no SQLite needed) ──────────────────────────
-
-function loadSources() {
-  if (!existsSync(SOURCES_FILE)) return [];
-  return JSON.parse(readFileSync(SOURCES_FILE, "utf-8"));
-}
-
-function saveSources(sources) {
-  writeFileSync(SOURCES_FILE, JSON.stringify(sources, null, 2));
-}
-
-function loadArticles() {
-  if (!existsSync(ARTICLES_FILE)) return {};
-  return JSON.parse(readFileSync(ARTICLES_FILE, "utf-8"));
-}
-
-function saveArticles(articles) {
-  writeFileSync(ARTICLES_FILE, JSON.stringify(articles, null, 2));
-}
-
-// ── RSS/Atom parser ───────────────────────────────────────────────────────────
-
-function extractTag(xml, tag) {
-  const m = xml.match(new RegExp(\`<\${tag}[^>]*>\\\\s*(?:<!\\\\[CDATA\\\\[)?([\\\\s\\\\S]*?)(?:\\\\]\\\\]>)?\\\\s*</\${tag}>\`, "i"));
-  return m ? m[1].trim() : "";
-}
-
-function extractAtomLink(xml) {
-  const alt = xml.match(/<link[^>]*rel="alternate"[^>]*href="([^"]+)"/i);
-  if (alt) return alt[1];
-  const any = xml.match(/<link[^>]*href="([^"]+)"/i);
-  return any ? any[1] : "";
-}
-
-function stripHtml(html) {
-  return html.replace(/<[^>]+>/g, "").replace(/&amp;/g, "&").replace(/&lt;/g, "<")
-    .replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&#39;/g, "'")
-    .replace(/\\s+/g, " ").trim();
-}
-
-async function parseFeed(url) {
-  const res = await fetch(url, {
-    headers: { "User-Agent": "${projectName} RSS Reader" },
-    signal: AbortSignal.timeout(15000),
-  });
-  if (!res.ok) throw new Error(\`Feed fetch failed: \${res.status}\`);
-  const xml = await res.text();
-  const items = [];
-
-  const rssItems = xml.match(/<item>([\\s\\S]*?)<\\/item>/gi) || [];
-  if (rssItems.length > 0) {
-    for (const item of rssItems) {
-      items.push({
-        title: stripHtml(extractTag(item, "title")),
-        link: extractTag(item, "link"),
-        published: extractTag(item, "pubDate"),
-        summary: stripHtml(extractTag(item, "description")).substring(0, 500),
-      });
-    }
-    return { title: stripHtml(extractTag(xml, "title")), items };
-  }
-
-  const atomEntries = xml.match(/<entry>([\\s\\S]*?)<\\/entry>/gi) || [];
-  for (const entry of atomEntries) {
-    items.push({
-      title: stripHtml(extractTag(entry, "title")),
-      link: extractAtomLink(entry) || extractTag(entry, "link"),
-      published: extractTag(entry, "published") || extractTag(entry, "updated"),
-      summary: stripHtml(extractTag(entry, "summary") || extractTag(entry, "content")).substring(0, 500),
-    });
-  }
-  return { title: stripHtml(extractTag(xml, "title")), items };
-}
-
-// ── SMTP email sender ─────────────────────────────────────────────────────────
-
-function readSmtp(socket, timeout = 10000) {
-  return new Promise((resolve, reject) => {
-    let buf = "";
-    const timer = setTimeout(() => { socket.removeAllListeners("data"); reject(new Error("SMTP timeout")); }, timeout);
-    const onData = (chunk) => {
-      buf += chunk.toString();
-      const lines = buf.split("\\r\\n").filter(Boolean);
-      const last = lines[lines.length - 1];
-      if (last && /^\\d{3} /.test(last)) {
-        clearTimeout(timer); socket.removeListener("data", onData);
-        const code = parseInt(last.substring(0, 3));
-        if (code >= 400) reject(new Error(\`SMTP \${code}: \${buf.trim()}\`));
-        else resolve(buf.trim());
-      }
-    };
-    socket.on("data", onData);
-  });
-}
-
-async function smtpCmd(socket, cmd) { socket.write(cmd + "\\r\\n"); return readSmtp(socket); }
-
-async function sendDigestEmail(to, subject, html, plainText) {
-  if (!EMAIL_USER || !EMAIL_PASS) {
-    console.log("  [skip] EMAIL_USER/EMAIL_PASS not set — digest printed to stdout instead");
-    console.log(plainText);
-    return;
-  }
-
-  const boundary = "----DigestBoundary" + Date.now();
-  const message = [
-    \`From: \${EMAIL_USER}\`, \`To: \${to}\`, \`Subject: \${subject}\`,
-    \`Date: \${new Date().toUTCString()}\`, "MIME-Version: 1.0",
-    \`Content-Type: multipart/alternative; boundary="\${boundary}"\`,
-    "", \`--\${boundary}\`, "Content-Type: text/plain; charset=UTF-8", "", plainText,
-    \`--\${boundary}\`, "Content-Type: text/html; charset=UTF-8", "", html,
-    \`--\${boundary}--\`,
-  ].join("\\r\\n");
-
-  const socket = tls.connect({ host: SMTP_HOST, port: SMTP_PORT, rejectUnauthorized: true });
-  await new Promise((resolve, reject) => { socket.once("secureConnect", resolve); socket.once("error", reject); });
-  try {
-    await readSmtp(socket);
-    await smtpCmd(socket, "EHLO digest-pipeline");
-    await smtpCmd(socket, "AUTH LOGIN");
-    await smtpCmd(socket, Buffer.from(EMAIL_USER).toString("base64"));
-    await smtpCmd(socket, Buffer.from(EMAIL_PASS).toString("base64"));
-    await smtpCmd(socket, \`MAIL FROM:<\${EMAIL_USER}>\`);
-    await smtpCmd(socket, \`RCPT TO:<\${to}>\`);
-    await smtpCmd(socket, "DATA");
-    socket.write(message.replace(/\\r\\n\\./g, "\\r\\n..") + "\\r\\n.\\r\\n");
-    await readSmtp(socket);
-    await smtpCmd(socket, "QUIT").catch(() => {});
-    console.log(\`  [sent] Digest emailed to \${to}\`);
-  } finally { socket.destroy(); }
-}
-
-// ── Main pipeline ─────────────────────────────────────────────────────────────
-
-async function fetchAll() {
-  const sources = loadSources();
-  if (sources.length === 0) {
-    console.log("No sources registered. Use: node digest.mjs --add-feed <url> [name] [category]");
-    return [];
-  }
-
-  const seenArticles = loadArticles();
-  const newArticles = [];
-
-  for (const source of sources) {
-    try {
-      const feed = await parseFeed(source.url);
-      let newCount = 0;
-      for (const item of feed.items.slice(0, 20)) {
-        if (!item.link) continue;
-        const key = source.url + "|" + item.link;
-        if (seenArticles[key]) continue;
-        seenArticles[key] = { fetchedAt: new Date().toISOString(), seen: false };
-        newArticles.push({ ...item, sourceName: source.name, category: source.category });
-        newCount++;
-      }
-      console.log(\`  [\${source.name}] \${feed.items.length} articles, \${newCount} new\`);
-    } catch (e) {
-      console.log(\`  [\${source.name}] ERROR: \${e.message}\`);
-    }
-  }
-
-  saveArticles(seenArticles);
-  return newArticles;
-}
-
-function buildDigest(articles) {
-  if (articles.length === 0) return { html: "", plainText: "", count: 0 };
-
-  const byCategory = new Map();
-  for (const a of articles) {
-    const cat = a.category || "general";
-    if (!byCategory.has(cat)) byCategory.set(cat, []);
-    byCategory.get(cat).push(a);
-  }
-
-  // Plain text
-  const lines = [\`Research Digest — \${articles.length} new articles — \${new Date().toLocaleDateString()}\`, ""];
-  for (const [cat, items] of byCategory) {
-    lines.push(\`## \${cat} (\${items.length})\`, "");
-    for (const a of items) {
-      lines.push(\`- \${a.title}\`);
-      lines.push(\`  \${a.sourceName} · \${a.published || "recent"}\`);
-      lines.push(\`  \${a.link}\`);
-      if (a.summary) lines.push(\`  > \${a.summary.substring(0, 200)}\`);
-      lines.push("");
-    }
-  }
-
-  // HTML
-  const sections = [...byCategory.entries()].map(([cat, items]) => \`
-    <h2 style="color:#333;border-bottom:2px solid #007bff;padding-bottom:4px">\${cat} (\${items.length})</h2>
-    \${items.map(a => \`
-      <div style="margin-bottom:16px;padding:12px;background:#f8f9fa;border-radius:6px">
-        <a href="\${a.link}" style="font-size:16px;font-weight:600;color:#007bff;text-decoration:none">\${a.title}</a>
-        <div style="color:#666;font-size:12px;margin-top:4px">\${a.sourceName} · \${a.published || "recent"}</div>
-        \${a.summary ? \`<p style="margin-top:8px;color:#444;font-size:14px">\${a.summary.substring(0, 300)}</p>\` : ""}
-      </div>\`).join("")}\`).join("");
-
-  const html = \`<div style="font-family:-apple-system,sans-serif;max-width:640px;margin:0 auto;padding:20px">
-    <h1 style="color:#111">Research Digest</h1>
-    <p style="color:#666">\${articles.length} new articles · \${new Date().toLocaleDateString()}</p>
-    \${sections}
-    <hr style="margin-top:32px;border:none;border-top:1px solid #ddd">
-    <p style="color:#999;font-size:11px">Generated by ${projectName}</p>
-  </div>\`;
-
-  return { html, plainText: lines.join("\\n"), count: articles.length };
-}
-
-// ── CLI ───────────────────────────────────────────────────────────────────────
-
-const args = process.argv.slice(2);
-
-if (args.includes("--list-sources")) {
-  const sources = loadSources();
-  if (sources.length === 0) console.log("No sources. Use --add-feed <url> [name] [category]");
-  else sources.forEach((s, i) => console.log(\`\${i + 1}. [\${s.category}] \${s.name} — \${s.url}\`));
-} else if (args.includes("--add-feed")) {
-  const idx = args.indexOf("--add-feed");
-  const url = args[idx + 1];
-  const name = args[idx + 2] || url;
-  const category = args[idx + 3] || "general";
-  if (!url) { console.error("Usage: --add-feed <url> [name] [category]"); process.exit(1); }
-  try {
-    const feed = await parseFeed(url);
-    const sources = loadSources();
-    sources.push({ url, name, category });
-    saveSources(sources);
-    console.log(\`Added: \${name} (\${feed.title}) — \${feed.items.length} articles available\`);
-  } catch (e) {
-    console.error(\`Failed to validate feed: \${e.message}\`);
-  }
-} else {
-  console.log(\`\\n${"=".repeat(60)}\`);
-  console.log(\`  ${projectName} — \${new Date().toISOString()}\`);
-  console.log(\`${"=".repeat(60)}\\n\`);
-
-  console.log("Fetching feeds...");
-  const articles = await fetchAll();
-
-  if (articles.length === 0) {
-    console.log("\\nNo new articles found.");
-  } else {
-    const digest = buildDigest(articles);
-    console.log(\`\\nDigest: \${digest.count} new articles\`);
-
-    if (!args.includes("--fetch-only")) {
-      const subject = \`Research Digest — \${digest.count} articles — \${new Date().toLocaleDateString()}\`;
-      await sendDigestEmail(EMAIL_TO, subject, digest.html, digest.plainText);
-    }
-
-    // Save digest to file
-    const digestFile = DATA_DIR + \`digest-\${new Date().toISOString().slice(0,10)}.md\`;
-    writeFileSync(digestFile, digest.plainText);
-    console.log(\`  [saved] \${digestFile}\`);
-  }
-}
+            const mainScript = `#!/usr/bin/env node
+/**
+ * ${projectName} — Automated Research Digest Pipeline
+ *
+ * Subscribe to RSS/Atom feeds, fetch new articles, build a digest, and email it.
+ * Generated by NodeBench MCP scaffold_research_pipeline.
+ *
+ * Usage:
+ *   node digest.mjs                    # Run once (fetch + digest + email)
+ *   node digest.mjs --fetch-only       # Just fetch, don't email
+ *   node digest.mjs --list-sources     # Show registered feeds
+ *   node digest.mjs --add-feed <url>   # Add a new feed
+ *
+ * Schedule with cron:
+ *   crontab -e
+ *   ${cronExpr[schedule]} cd /path/to/${projectName} && node digest.mjs >> digest.log 2>&1
+ */
+
+import { readFileSync, writeFileSync, existsSync, mkdirSync } from "node:fs";
+import * as tls from "node:tls";
+
+// ── Config ────────────────────────────────────────────────────────────────────
+
+const DATA_DIR = new URL("./data/", import.meta.url).pathname.replace(/^\\//, "");
+if (!existsSync(DATA_DIR)) mkdirSync(DATA_DIR, { recursive: true });
+
+const SOURCES_FILE = DATA_DIR + "sources.json";
+const ARTICLES_FILE = DATA_DIR + "articles.json";
+
+const EMAIL_USER = process.env.EMAIL_USER || "";
+const EMAIL_PASS = process.env.EMAIL_PASS || "";
+const EMAIL_TO = process.env.DIGEST_TO || "${emailTo}" || EMAIL_USER;
+const SMTP_HOST = process.env.EMAIL_SMTP_HOST || "smtp.gmail.com";
+const SMTP_PORT = parseInt(process.env.EMAIL_SMTP_PORT || "465");
+
+// ── Feed storage (JSON file-based, no SQLite needed) ──────────────────────────
+
+function loadSources() {
+  if (!existsSync(SOURCES_FILE)) return [];
+  return JSON.parse(readFileSync(SOURCES_FILE, "utf-8"));
+}
+
+function saveSources(sources) {
+  writeFileSync(SOURCES_FILE, JSON.stringify(sources, null, 2));
+}
+
+function loadArticles() {
+  if (!existsSync(ARTICLES_FILE)) return {};
+  return JSON.parse(readFileSync(ARTICLES_FILE, "utf-8"));
+}
+
+function saveArticles(articles) {
+  writeFileSync(ARTICLES_FILE, JSON.stringify(articles, null, 2));
+}
+
+// ── RSS/Atom parser ───────────────────────────────────────────────────────────
+
+function extractTag(xml, tag) {
+  const m = xml.match(new RegExp(\`<\${tag}[^>]*>\\\\s*(?:<!\\\\[CDATA\\\\[)?([\\\\s\\\\S]*?)(?:\\\\]\\\\]>)?\\\\s*</\${tag}>\`, "i"));
+  return m ? m[1].trim() : "";
+}
+
+function extractAtomLink(xml) {
+  const alt = xml.match(/<link[^>]*rel="alternate"[^>]*href="([^"]+)"/i);
+  if (alt) return alt[1];
+  const any = xml.match(/<link[^>]*href="([^"]+)"/i);
+  return any ? any[1] : "";
+}
+
+function stripHtml(html) {
+  return html.replace(/<[^>]+>/g, "").replace(/&amp;/g, "&").replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">").replace(/&quot;/g, '"').replace(/&#39;/g, "'")
+    .replace(/\\s+/g, " ").trim();
+}
+
+async function parseFeed(url) {
+  const res = await fetch(url, {
+    headers: { "User-Agent": "${projectName} RSS Reader" },
+    signal: AbortSignal.timeout(15000),
+  });
+  if (!res.ok) throw new Error(\`Feed fetch failed: \${res.status}\`);
+  const xml = await res.text();
+  const items = [];
+
+  const rssItems = xml.match(/<item>([\\s\\S]*?)<\\/item>/gi) || [];
+  if (rssItems.length > 0) {
+    for (const item of rssItems) {
+      items.push({
+        title: stripHtml(extractTag(item, "title")),
+        link: extractTag(item, "link"),
+        published: extractTag(item, "pubDate"),
+        summary: stripHtml(extractTag(item, "description")).substring(0, 500),
+      });
+    }
+    return { title: stripHtml(extractTag(xml, "title")), items };
+  }
+
+  const atomEntries = xml.match(/<entry>([\\s\\S]*?)<\\/entry>/gi) || [];
+  for (const entry of atomEntries) {
+    items.push({
+      title: stripHtml(extractTag(entry, "title")),
+      link: extractAtomLink(entry) || extractTag(entry, "link"),
+      published: extractTag(entry, "published") || extractTag(entry, "updated"),
+      summary: stripHtml(extractTag(entry, "summary") || extractTag(entry, "content")).substring(0, 500),
+    });
+  }
+  return { title: stripHtml(extractTag(xml, "title")), items };
+}
+
+// ── SMTP email sender ─────────────────────────────────────────────────────────
+
+function readSmtp(socket, timeout = 10000) {
+  return new Promise((resolve, reject) => {
+    let buf = "";
+    const timer = setTimeout(() => { socket.removeAllListeners("data"); reject(new Error("SMTP timeout")); }, timeout);
+    const onData = (chunk) => {
+      buf += chunk.toString();
+      const lines = buf.split("\\r\\n").filter(Boolean);
+      const last = lines[lines.length - 1];
+      if (last && /^\\d{3} /.test(last)) {
+        clearTimeout(timer); socket.removeListener("data", onData);
+        const code = parseInt(last.substring(0, 3));
+        if (code >= 400) reject(new Error(\`SMTP \${code}: \${buf.trim()}\`));
+        else resolve(buf.trim());
+      }
+    };
+    socket.on("data", onData);
+  });
+}
+
+async function smtpCmd(socket, cmd) { socket.write(cmd + "\\r\\n"); return readSmtp(socket); }
+
+async function sendDigestEmail(to, subject, html, plainText) {
+  if (!EMAIL_USER || !EMAIL_PASS) {
+    console.log("  [skip] EMAIL_USER/EMAIL_PASS not set — digest printed to stdout instead");
+    console.log(plainText);
+    return;
+  }
+
+  const boundary = "----DigestBoundary" + Date.now();
+  const message = [
+    \`From: \${EMAIL_USER}\`, \`To: \${to}\`, \`Subject: \${subject}\`,
+    \`Date: \${new Date().toUTCString()}\`, "MIME-Version: 1.0",
+    \`Content-Type: multipart/alternative; boundary="\${boundary}"\`,
+    "", \`--\${boundary}\`, "Content-Type: text/plain; charset=UTF-8", "", plainText,
+    \`--\${boundary}\`, "Content-Type: text/html; charset=UTF-8", "", html,
+    \`--\${boundary}--\`,
+  ].join("\\r\\n");
+
+  const socket = tls.connect({ host: SMTP_HOST, port: SMTP_PORT, rejectUnauthorized: true });
+  await new Promise((resolve, reject) => { socket.once("secureConnect", resolve); socket.once("error", reject); });
+  try {
+    await readSmtp(socket);
+    await smtpCmd(socket, "EHLO digest-pipeline");
+    await smtpCmd(socket, "AUTH LOGIN");
+    await smtpCmd(socket, Buffer.from(EMAIL_USER).toString("base64"));
+    await smtpCmd(socket, Buffer.from(EMAIL_PASS).toString("base64"));
+    await smtpCmd(socket, \`MAIL FROM:<\${EMAIL_USER}>\`);
+    await smtpCmd(socket, \`RCPT TO:<\${to}>\`);
+    await smtpCmd(socket, "DATA");
+    socket.write(message.replace(/\\r\\n\\./g, "\\r\\n..") + "\\r\\n.\\r\\n");
+    await readSmtp(socket);
+    await smtpCmd(socket, "QUIT").catch(() => {});
+    console.log(\`  [sent] Digest emailed to \${to}\`);
+  } finally { socket.destroy(); }
+}
+
+// ── Main pipeline ─────────────────────────────────────────────────────────────
+
+async function fetchAll() {
+  const sources = loadSources();
+  if (sources.length === 0) {
+    console.log("No sources registered. Use: node digest.mjs --add-feed <url> [name] [category]");
+    return [];
+  }
+
+  const seenArticles = loadArticles();
+  const newArticles = [];
+
+  for (const source of sources) {
+    try {
+      const feed = await parseFeed(source.url);
+      let newCount = 0;
+      for (const item of feed.items.slice(0, 20)) {
+        if (!item.link) continue;
+        const key = source.url + "|" + item.link;
+        if (seenArticles[key]) continue;
+        seenArticles[key] = { fetchedAt: new Date().toISOString(), seen: false };
+        newArticles.push({ ...item, sourceName: source.name, category: source.category });
+        newCount++;
+      }
+      console.log(\`  [\${source.name}] \${feed.items.length} articles, \${newCount} new\`);
+    } catch (e) {
+      console.log(\`  [\${source.name}] ERROR: \${e.message}\`);
+    }
+  }
+
+  saveArticles(seenArticles);
+  return newArticles;
+}
+
+function buildDigest(articles) {
+  if (articles.length === 0) return { html: "", plainText: "", count: 0 };
+
+  const byCategory = new Map();
+  for (const a of articles) {
+    const cat = a.category || "general";
+    if (!byCategory.has(cat)) byCategory.set(cat, []);
+    byCategory.get(cat).push(a);
+  }
+
+  // Plain text
+  const lines = [\`Research Digest — \${articles.length} new articles — \${new Date().toLocaleDateString()}\`, ""];
+  for (const [cat, items] of byCategory) {
+    lines.push(\`## \${cat} (\${items.length})\`, "");
+    for (const a of items) {
+      lines.push(\`- \${a.title}\`);
+      lines.push(\`  \${a.sourceName} · \${a.published || "recent"}\`);
+      lines.push(\`  \${a.link}\`);
+      if (a.summary) lines.push(\`  > \${a.summary.substring(0, 200)}\`);
+      lines.push("");
+    }
+  }
+
+  // HTML
+  const sections = [...byCategory.entries()].map(([cat, items]) => \`
+    <h2 style="color:#333;border-bottom:2px solid #007bff;padding-bottom:4px">\${cat} (\${items.length})</h2>
+    \${items.map(a => \`
+      <div style="margin-bottom:16px;padding:12px;background:#f8f9fa;border-radius:6px">
+        <a href="\${a.link}" style="font-size:16px;font-weight:600;color:#007bff;text-decoration:none">\${a.title}</a>
+        <div style="color:#666;font-size:12px;margin-top:4px">\${a.sourceName} · \${a.published || "recent"}</div>
+        \${a.summary ? \`<p style="margin-top:8px;color:#444;font-size:14px">\${a.summary.substring(0, 300)}</p>\` : ""}
+      </div>\`).join("")}\`).join("");
+
+  const html = \`<div style="font-family:-apple-system,sans-serif;max-width:640px;margin:0 auto;padding:20px">
+    <h1 style="color:#111">Research Digest</h1>
+    <p style="color:#666">\${articles.length} new articles · \${new Date().toLocaleDateString()}</p>
+    \${sections}
+    <hr style="margin-top:32px;border:none;border-top:1px solid #ddd">
+    <p style="color:#999;font-size:11px">Generated by ${projectName}</p>
+  </div>\`;
+
+  return { html, plainText: lines.join("\\n"), count: articles.length };
+}
+
+// ── CLI ───────────────────────────────────────────────────────────────────────
+
+const args = process.argv.slice(2);
+
+if (args.includes("--list-sources")) {
+  const sources = loadSources();
+  if (sources.length === 0) console.log("No sources. Use --add-feed <url> [name] [category]");
+  else sources.forEach((s, i) => console.log(\`\${i + 1}. [\${s.category}] \${s.name} — \${s.url}\`));
+} else if (args.includes("--add-feed")) {
+  const idx = args.indexOf("--add-feed");
+  const url = args[idx + 1];
+  const name = args[idx + 2] || url;
+  const category = args[idx + 3] || "general";
+  if (!url) { console.error("Usage: --add-feed <url> [name] [category]"); process.exit(1); }
+  try {
+    const feed = await parseFeed(url);
+    const sources = loadSources();
+    sources.push({ url, name, category });
+    saveSources(sources);
+    console.log(\`Added: \${name} (\${feed.title}) — \${feed.items.length} articles available\`);
+  } catch (e) {
+    console.error(\`Failed to validate feed: \${e.message}\`);
+  }
+} else {
+  console.log(\`\\n${"=".repeat(60)}\`);
+  console.log(\`  ${projectName} — \${new Date().toISOString()}\`);
+  console.log(\`${"=".repeat(60)}\\n\`);
+
+  console.log("Fetching feeds...");
+  const articles = await fetchAll();
+
+  if (articles.length === 0) {
+    console.log("\\nNo new articles found.");
+  } else {
+    const digest = buildDigest(articles);
+    console.log(\`\\nDigest: \${digest.count} new articles\`);
+
+    if (!args.includes("--fetch-only")) {
+      const subject = \`Research Digest — \${digest.count} articles — \${new Date().toLocaleDateString()}\`;
+      await sendDigestEmail(EMAIL_TO, subject, digest.html, digest.plainText);
+    }
+
+    // Save digest to file
+    const digestFile = DATA_DIR + \`digest-\${new Date().toISOString().slice(0,10)}.md\`;
+    writeFileSync(digestFile, digest.plainText);
+    console.log(\`  [saved] \${digestFile}\`);
+  }
+}
 `;
             // ── Generate package.json ──
             const packageJson = JSON.stringify({
@@ -737,67 +737,67 @@ if (args.includes("--list-sources")) {
                 license: "MIT",
             }, null, 2);
             // ── Generate .env template ──
-            const envTemplate = `# Email configuration (required for email delivery)
-# For Gmail: use an App Password (Google Account → Security → App passwords)
-EMAIL_USER=your.email@gmail.com
-EMAIL_PASS=your-16-char-app-password
-DIGEST_TO=${emailTo || "your.email@gmail.com"}
-
-# Optional: non-Gmail SMTP
-# EMAIL_SMTP_HOST=smtp.gmail.com
-# EMAIL_SMTP_PORT=465
+            const envTemplate = `# Email configuration (required for email delivery)
+# For Gmail: use an App Password (Google Account → Security → App passwords)
+EMAIL_USER=your.email@gmail.com
+EMAIL_PASS=your-16-char-app-password
+DIGEST_TO=${emailTo || "your.email@gmail.com"}
+
+# Optional: non-Gmail SMTP
+# EMAIL_SMTP_HOST=smtp.gmail.com
+# EMAIL_SMTP_PORT=465
 `;
             // ── Generate README ──
-            const readme = `# ${projectName}
-
-Automated research digest pipeline. Subscribes to RSS/Atom feeds, fetches new articles, builds a categorized digest, and emails it to you.
-
-## Quick Start
-
-\`\`\`bash
-# 1. Set up email (see .env.example)
-cp .env.example .env
-# Edit .env with your email credentials
-
-# 2. Add feeds
-node digest.mjs --add-feed "https://arxiv.org/rss/cs.AI" "arXiv AI" "ai-research"
-node digest.mjs --add-feed "https://hnrss.org/newest?points=100" "HN Top" "tech"
-node digest.mjs --add-feed "https://blog.anthropic.com/rss.xml" "Anthropic" "ai-research"
-
-# 3. Run
-node digest.mjs
-\`\`\`
-
-## Schedule (cron)
-
-\`\`\`bash
-# Edit crontab
-crontab -e
-
-# Add (${schedule}):
-${cronExpr[schedule]} cd /path/to/${projectName} && node digest.mjs >> digest.log 2>&1
-\`\`\`
-
-## Commands
-
-| Command | What it does |
-|---|---|
-| \`node digest.mjs\` | Fetch + digest + email |
-| \`node digest.mjs --fetch-only\` | Fetch only (no email) |
-| \`node digest.mjs --list-sources\` | Show registered feeds |
-| \`node digest.mjs --add-feed <url> [name] [category]\` | Add a new feed |
-
-## How It Works
-
-1. **Fetch**: Pulls latest articles from all registered RSS/Atom feeds
-2. **Deduplicate**: Skips articles already seen (tracked in \`data/articles.json\`)
-3. **Digest**: Builds a categorized summary (plain text + HTML)
-4. **Email**: Sends via SMTP over TLS (Gmail default, configurable)
-5. **Save**: Archives digest as markdown in \`data/\`
-
-No dependencies. Pure Node.js (>= 18). Zero npm packages.
-
-Generated by [NodeBench MCP](https://www.npmjs.com/package/nodebench-mcp) \`scaffold_research_pipeline\`
+            const readme = `# ${projectName}
+
+Automated research digest pipeline. Subscribes to RSS/Atom feeds, fetches new articles, builds a categorized digest, and emails it to you.
+
+## Quick Start
+
+\`\`\`bash
+# 1. Set up email (see .env.example)
+cp .env.example .env
+# Edit .env with your email credentials
+
+# 2. Add feeds
+node digest.mjs --add-feed "https://arxiv.org/rss/cs.AI" "arXiv AI" "ai-research"
+node digest.mjs --add-feed "https://hnrss.org/newest?points=100" "HN Top" "tech"
+node digest.mjs --add-feed "https://blog.anthropic.com/rss.xml" "Anthropic" "ai-research"
+
+# 3. Run
+node digest.mjs
+\`\`\`
+
+## Schedule (cron)
+
+\`\`\`bash
+# Edit crontab
+crontab -e
+
+# Add (${schedule}):
+${cronExpr[schedule]} cd /path/to/${projectName} && node digest.mjs >> digest.log 2>&1
+\`\`\`
+
+## Commands
+
+| Command | What it does |
+|---|---|
+| \`node digest.mjs\` | Fetch + digest + email |
+| \`node digest.mjs --fetch-only\` | Fetch only (no email) |
+| \`node digest.mjs --list-sources\` | Show registered feeds |
+| \`node digest.mjs --add-feed <url> [name] [category]\` | Add a new feed |
+
+## How It Works
+
+1. **Fetch**: Pulls latest articles from all registered RSS/Atom feeds
+2. **Deduplicate**: Skips articles already seen (tracked in \`data/articles.json\`)
+3. **Digest**: Builds a categorized summary (plain text + HTML)
+4. **Email**: Sends via SMTP over TLS (Gmail default, configurable)
+5. **Save**: Archives digest as markdown in \`data/\`
+
+No dependencies. Pure Node.js (>= 18). Zero npm packages.
+
+Generated by [NodeBench MCP](https://www.npmjs.com/package/nodebench-mcp) \`scaffold_research_pipeline\`
 `;
             return [
                 {