@mdvp/cli 1.9.1 → 1.11.0

package/cli.mjs

@@ -299,6 +299,9 @@ ${BOLD}Account${R}
   submit <domain> --local     Submit to local crawler node instead
 
 ${BOLD}Crawler${R}
+  recrawl                     Re-queue existing sites for recrawl (updates data)
+  recrawl linear.app          Re-queue specific sites
+  recrawl --limit=100         Re-queue oldest N sites
   hire                        Become a crawler node (downloads + runs worker)
   hire --daemon               Run crawler in background
   hire --tabs=4               Run with 4 parallel tabs
@@ -357,8 +360,15 @@ async function cmdHire(opts) {
   await download(extractUrl, `${dir}/extract.js`)
     .catch(() => {})
 
-  if (!existsSync(`${dir}/package.json`)) {
-    writeFileSync(`${dir}/package.json`, '{"type":"module"}')
+  writeFileSync(`${dir}/package.json`, '{"type":"module","dependencies":{"puppeteer":"*"}}')
+
+  const needsInstall = !existsSync(`${dir}/node_modules/puppeteer`)
+  if (needsInstall) {
+    process.stderr.write(`${DIM}Installing puppeteer (first run, ~30s)...${R}\n`)
+    await new Promise((res, rej) => {
+      const child = spawn("npm", ["install", "--prefer-offline"], { cwd: dir, stdio: "inherit" })
+      child.on("exit", (code) => code === 0 ? res() : rej(new Error(`npm install failed (${code})`)))
+    })
   }
 
   const apiUrl = local ? "http://localhost:7227" : API
@@ -391,6 +401,20 @@ async function cmdHire(opts) {
   }
 }
 
+async function cmdRecrawl(opts, domains) {
+  const { apiKey } = opts
+  if (!apiKey) { console.error(`${RED}No API key. Run: npx @mdvp/cli login${R}`); process.exit(1) }
+
+  const limit = opts.limit || 50
+  const body = domains.length > 0 ? { domains } : { limit }
+  process.stderr.write(`${DIM}queuing ${domains.length > 0 ? domains.join(', ') : `up to ${limit} oldest`} for recrawl...${R}\n`)
+
+  const d = await apiPost("/crawl/recrawl", body, apiKey)
+  if (d.error) { console.error(`${RED}Error: ${d.error}${R}`); process.exit(1) }
+  console.log(`\n  ${BOLD}Queued!${R} ${d.queued} sites scheduled for recrawl.`)
+  console.log(`  ${DIM}Run: npx @mdvp/cli hire --tabs=4 to start crawling${R}\n`)
+}
+
 async function cmdSubmit(domain, opts) {
   const { apiKey, local } = opts
   domain = parseDomain(domain)
@@ -452,6 +476,9 @@ async function main() {
     const d = await apiGet("/dataset/stats")
     if (opts.json) { console.log(JSON.stringify(d, null, 2)); return }
     console.log(`\n  Total sites:   ${d.totalSites}\n  Average score: ${d.averageScore}\n`)
+  } else if (cmd === "recrawl") {
+    const domains = positional.slice(1)  // optional: npx @mdvp/cli recrawl linear.app vercel.com
+    await cmdRecrawl({ ...opts, limit: parseInt(flagValues.limit || "50") || 50 }, domains)
   } else if (cmd === "hire" || cmd === "apply" || cmd === "serve") {
     await cmdHire(opts)
   } else if (cmd === "label" && arg1) {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mdvp/cli",
-  "version": "1.9.1",
+  "version": "1.11.0",
   "description": "Machine Design Vision Protocol — gives agents eyes to understand design quality",
   "bin": {
     "mdvp": "./cli.mjs"