diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index da6d8b2b..335ad126 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -169,6 +169,10 @@ jobs:
     if: needs.checks.result == 'success'
     # pull → local `vercel build` (pnpm install + Next/Payload build against prod env) → upload.
     timeout-minutes: 35
+    outputs:
+      stale: ${{ steps.tip.outputs.stale || 'false' }}
+      cooldown: ${{ steps.cooldown.outputs.active || 'false' }}
+      wait_seconds: ${{ steps.cooldown.outputs.wait_seconds || '0' }}
     env:
       VERCEL_ORG_ID: ${{ secrets.VERCEL_ORG_ID }}
       VERCEL_PROJECT_ID: ${{ secrets.VERCEL_PROJECT_ID }}
@@ -189,6 +193,10 @@ jobs:
             exit 0
           fi
           echo "stale=false" >> "$GITHUB_OUTPUT"
+      - uses: actions/setup-node@v5
+        if: steps.tip.outputs.stale != 'true'
+        with:
+          node-version: 24
       - name: Require Vercel credentials
         if: steps.tip.outputs.stale != 'true'
         run: |
@@ -212,20 +220,29 @@ jobs:
             exit 1
             ;;
           esac
+      - name: Check deploy cooldown
+        id: cooldown
+        if: steps.tip.outputs.stale != 'true'
+        run: node scripts/vercel-deploy-cooldown.mjs
+      - name: Log deploy deferral
+        if: steps.tip.outputs.stale != 'true' && steps.cooldown.outputs.active == 'true'
+        run: |
+          echo "Deploy deferred: last production deploy is younger than 30 minutes."
+          echo "Requeue will retry in ${{ steps.cooldown.outputs.wait_seconds }}s."
       # `vercel build` shells out to pnpm (packageManager in package.json); without
       # these steps the runner only has npm → spawn pnpm ENOENT.
       - uses: pnpm/action-setup@v6
-        if: steps.tip.outputs.stale != 'true'
+        if: steps.tip.outputs.stale != 'true' && steps.cooldown.outputs.active != 'true'
       - uses: actions/setup-node@v5
-        if: steps.tip.outputs.stale != 'true'
+        if: steps.tip.outputs.stale != 'true' && steps.cooldown.outputs.active != 'true'
         with:
           node-version: 24
           cache: pnpm
       - name: Install Vercel CLI
-        if: steps.tip.outputs.stale != 'true'
+        if: steps.tip.outputs.stale != 'true' && steps.cooldown.outputs.active != 'true'
         run: npm install --global vercel@55
       - name: Deploy production
-        if: steps.tip.outputs.stale != 'true'
+        if: steps.tip.outputs.stale != 'true' && steps.cooldown.outputs.active != 'true'
         run: |
           set -euo pipefail
           # Workflow-level env points at the CI teqo_test service. Drop it so
@@ -236,12 +253,13 @@ jobs:
           # "Could not retrieve Project Settings" usually means bad/expired token or wrong IDs (403).
           vercel pull --yes --environment=production --token="$VERCEL_TOKEN"
           vercel build --prod --token="$VERCEL_TOKEN"
-          vercel deploy --prebuilt --prod --token="$VERCEL_TOKEN"
+          vercel deploy --prebuilt --prod --archive=tgz --token="$VERCEL_TOKEN"
 
   requeue:
     runs-on: ubuntu-latest
     needs: [checks, deploy]
     if: always()
+    timeout-minutes: 40
     permissions:
       actions: write
       contents: read
@@ -249,17 +267,38 @@ jobs:
       - uses: actions/checkout@v5
         with:
           fetch-depth: 0
-      - name: Re-dispatch if main advanced
+      - name: Re-dispatch if main advanced or deploy deferred
         env:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           GH_REPO: ${{ github.repository }}
+          DEPLOY_RESULT: ${{ needs.deploy.result }}
+          DEPLOY_COOLDOWN: ${{ needs.deploy.outputs.cooldown }}
+          DEPLOY_WAIT_SECONDS: ${{ needs.deploy.outputs.wait_seconds }}
         run: |
           set -euo pipefail
           git fetch origin main --quiet
           tip=$(git rev-parse origin/main)
-          if [ "$tip" = "$GITHUB_SHA" ]; then
-            echo "HEAD still $GITHUB_SHA — idle"
+          if [ "$tip" != "$GITHUB_SHA" ]; then
+            echo "origin/main moved to $tip (verified $GITHUB_SHA) — requeue ci.yml"
+            gh workflow run ci.yml --ref main
             exit 0
           fi
-          echo "origin/main moved to $tip (verified $GITHUB_SHA) — requeue ci.yml"
-          gh workflow run ci.yml --ref main
+          if [ "${DEPLOY_RESULT:-}" != "success" ]; then
+            echo "Deploy job did not succeed (${DEPLOY_RESULT:-skipped}) — idle"
+            exit 0
+          fi
+          if [ "${DEPLOY_COOLDOWN:-}" = "true" ]; then
+            wait="${DEPLOY_WAIT_SECONDS:-0}"
+            echo "Deploy deferred by cooldown — waiting ${wait}s then requeue ci.yml"
+            sleep "$wait"
+            git fetch origin main --quiet
+            tip=$(git rev-parse origin/main)
+            if [ "$tip" != "$GITHUB_SHA" ]; then
+              echo "main advanced during cooldown wait ($tip) — requeue for new tip"
+            else
+              echo "Cooldown elapsed for $GITHUB_SHA — requeue ci.yml"
+            fi
+            gh workflow run ci.yml --ref main
+            exit 0
+          fi
+          echo "HEAD still $GITHUB_SHA — idle"
diff --git a/scripts/lib/vercel-deploy-cooldown.mjs b/scripts/lib/vercel-deploy-cooldown.mjs
new file mode 100644
index 00000000..86199037
--- /dev/null
+++ b/scripts/lib/vercel-deploy-cooldown.mjs
@@ -0,0 +1,122 @@
+/**
+ * Production deploy cooldown — queries Vercel for the last READY production
+ * deployment and computes whether CI should defer another upload (OPS11).
+ */
+
+export const DEPLOY_COOLDOWN_MS = 30 * 60 * 1000
+
+const VERCEL_API_BASE = 'https://api.vercel.com'
+
+/**
+ * @param {number | null | undefined} lastReadyAtMs
+ * @param {number} [nowMs]
+ */
+export const computeCooldownWait = (lastReadyAtMs, nowMs = Date.now()) => {
+  if (lastReadyAtMs == null || !Number.isFinite(lastReadyAtMs)) {
+    return { active: false, waitSeconds: 0, ageMs: null }
+  }
+
+  const ageMs = nowMs - lastReadyAtMs
+  const remainingMs = DEPLOY_COOLDOWN_MS - ageMs
+
+  if (remainingMs <= 0) {
+    return { active: false, waitSeconds: 0, ageMs }
+  }
+
+  return {
+    active: true,
+    waitSeconds: Math.ceil(remainingMs / 1000),
+    ageMs,
+  }
+}
+
+/**
+ * @param {unknown} deployment
+ * @returns {number | null}
+ */
+export const lastProductionReadyAtFromDeployment = (deployment) => {
+  if (!deployment || typeof deployment !== 'object') return null
+  const ready = /** @type {{ ready?: unknown; createdAt?: unknown }} */ (deployment).ready
+  const createdAt = /** @type {{ ready?: unknown; createdAt?: unknown }} */ (deployment).createdAt
+  const timestamp = ready ?? createdAt
+  return typeof timestamp === 'number' && Number.isFinite(timestamp) ? timestamp : null
+}
+
+/**
+ * @param {{
+ *   token: string
+ *   projectId: string
+ *   teamId: string
+ *   fetchImpl?: typeof fetch
+ * }} options
+ * @returns {Promise<number | null>}
+ */
+export const fetchLastProductionReadyAt = async ({
+  token,
+  projectId,
+  teamId,
+  fetchImpl = fetch,
+}) => {
+  const params = new URLSearchParams({
+    projectId,
+    teamId,
+    target: 'production',
+    state: 'READY',
+    limit: '1',
+  })
+
+  const response = await fetchImpl(`${VERCEL_API_BASE}/v6/deployments?${params}`, {
+    headers: { authorization: `Bearer ${token}` },
+  })
+
+  const bodyText = await response.text()
+  let body
+  try {
+    body = bodyText ? JSON.parse(bodyText) : {}
+  } catch {
+    throw new Error(`Vercel API returned non-JSON (${response.status})`)
+  }
+
+  if (!response.ok) {
+    const message =
+      typeof body?.error?.message === 'string'
+        ? body.error.message
+        : bodyText || `HTTP ${response.status}`
+    throw new Error(`Vercel deployments API ${response.status}: ${message}`)
+  }
+
+  const deployments = Array.isArray(body.deployments) ? body.deployments : []
+  return lastProductionReadyAtFromDeployment(deployments[0])
+}
+
+/**
+ * @param {{
+ *   token: string
+ *   projectId: string
+ *   teamId: string
+ *   fetchImpl?: typeof fetch
+ *   nowMs?: number
+ * }} options
+ */
+export const evaluateDeployCooldown = async ({
+  token,
+  projectId,
+  teamId,
+  fetchImpl = fetch,
+  nowMs = Date.now(),
+}) => {
+  const lastProductionDeployAt = await fetchLastProductionReadyAt({
+    token,
+    projectId,
+    teamId,
+    fetchImpl,
+  })
+  const { active, waitSeconds, ageMs } = computeCooldownWait(lastProductionDeployAt, nowMs)
+
+  return {
+    active,
+    waitSeconds,
+    ageMs,
+    lastProductionDeployAt,
+  }
+}
diff --git a/scripts/vercel-deploy-cooldown.mjs b/scripts/vercel-deploy-cooldown.mjs
new file mode 100644
index 00000000..fd99282f
--- /dev/null
+++ b/scripts/vercel-deploy-cooldown.mjs
@@ -0,0 +1,71 @@
+#!/usr/bin/env node
+/**
+ * vercel-deploy-cooldown — CI helper for OPS11. Queries the last READY
+ * production deployment on Vercel and writes GitHub Actions outputs:
+ *   active=true|false
+ *   wait_seconds=<residual>
+ *   last_production_deploy_at=<ms epoch or empty>
+ */
+import { appendFileSync } from 'node:fs'
+
+import { dieWithLabel } from './lib/cli.mjs'
+import {
+  DEPLOY_COOLDOWN_MS,
+  evaluateDeployCooldown,
+} from './lib/vercel-deploy-cooldown.mjs'
+
+const die = dieWithLabel('vercel-deploy-cooldown')
+
+const writeGithubOutput = (pairs) => {
+  const outputPath = process.env.GITHUB_OUTPUT
+  const lines = Object.entries(pairs)
+    .map(([key, value]) => `${key}=${value}`)
+    .join('\n')
+    .concat('\n')
+
+  if (outputPath) {
+    appendFileSync(outputPath, lines)
+    return
+  }
+
+  process.stdout.write(lines)
+}
+
+const main = async () => {
+  const token = process.env.VERCEL_TOKEN
+  const projectId = process.env.VERCEL_PROJECT_ID
+  const teamId = process.env.VERCEL_ORG_ID
+
+  if (!token) die('VERCEL_TOKEN is required')
+  if (!projectId) die('VERCEL_PROJECT_ID is required')
+  if (!teamId) die('VERCEL_ORG_ID is required')
+
+  const result = await evaluateDeployCooldown({ token, projectId, teamId })
+
+  if (result.lastProductionDeployAt == null) {
+    console.log(
+      `[vercel-deploy-cooldown] no READY production deployment found — proceed (${DEPLOY_COOLDOWN_MS / 60_000} min cooldown N/A)`,
+    )
+  } else if (result.active) {
+    const ageMinutes = Math.floor((result.ageMs ?? 0) / 60_000)
+    console.log(
+      `[vercel-deploy-cooldown] defer deploy — last production deploy ${ageMinutes} min ago; wait ${result.waitSeconds}s`,
+    )
+  } else {
+    const ageMinutes = Math.floor((result.ageMs ?? 0) / 60_000)
+    console.log(
+      `[vercel-deploy-cooldown] cooldown clear — last production deploy ${ageMinutes} min ago`,
+    )
+  }
+
+  writeGithubOutput({
+    active: result.active ? 'true' : 'false',
+    wait_seconds: String(result.waitSeconds),
+    last_production_deploy_at:
+      result.lastProductionDeployAt == null ? '' : String(result.lastProductionDeployAt),
+  })
+}
+
+main().catch((error) => {
+  die(error instanceof Error ? error.message : String(error))
+})
diff --git a/tests/unit/vercelDeployCooldown.unit.spec.ts b/tests/unit/vercelDeployCooldown.unit.spec.ts
new file mode 100644
index 00000000..e8e1bd11
--- /dev/null
+++ b/tests/unit/vercelDeployCooldown.unit.spec.ts
@@ -0,0 +1,107 @@
+// @vitest-environment node
+
+import { describe, expect, it } from 'vitest'
+
+import {
+  computeCooldownWait,
+  DEPLOY_COOLDOWN_MS,
+  evaluateDeployCooldown,
+  fetchLastProductionReadyAt,
+  lastProductionReadyAtFromDeployment,
+} from '../../scripts/lib/vercel-deploy-cooldown.mjs'
+
+const fakeResponse = (ok: boolean, status: number, payload: unknown) =>
+  ({
+    ok,
+    status,
+    text: async () => (payload === null ? '' : JSON.stringify(payload)),
+  }) as Response
+
+const captureFetch = (response: Response) => {
+  const calls: Array<{ url: string; init: RequestInit }> = []
+  const fetchImpl = async (url: string | URL | Request, init?: RequestInit) => {
+    calls.push({ url: String(url), init: init ?? {} })
+    return response
+  }
+  return { calls, fetchImpl: fetchImpl as typeof fetch }
+}
+
+describe('vercelDeployCooldown', () => {
+  it('uses a 30-minute cooldown constant', () => {
+    expect(DEPLOY_COOLDOWN_MS).toBe(30 * 60 * 1000)
+  })
+
+  it('lastProductionReadyAtFromDeployment prefers ready over createdAt', () => {
+    expect(lastProductionReadyAtFromDeployment({ ready: 100, createdAt: 50 })).toBe(100)
+    expect(lastProductionReadyAtFromDeployment({ createdAt: 50 })).toBe(50)
+    expect(lastProductionReadyAtFromDeployment(null)).toBeNull()
+  })
+
+  it('computeCooldownWait is inactive when no prior deploy exists', () => {
+    expect(computeCooldownWait(null, 1_000_000)).toEqual({
+      active: false,
+      waitSeconds: 0,
+      ageMs: null,
+    })
+  })
+
+  it('computeCooldownWait defers when the last deploy is younger than 30 minutes', () => {
+    const now = 1_700_000_000_000
+    const last = now - 10 * 60 * 1000
+
+    expect(computeCooldownWait(last, now)).toEqual({
+      active: true,
+      waitSeconds: 20 * 60,
+      ageMs: 10 * 60 * 1000,
+    })
+  })
+
+  it('computeCooldownWait clears when the last deploy is older than 30 minutes', () => {
+    const now = 1_700_000_000_000
+    const last = now - 31 * 60 * 1000
+
+    expect(computeCooldownWait(last, now)).toEqual({
+      active: false,
+      waitSeconds: 0,
+      ageMs: 31 * 60 * 1000,
+    })
+  })
+
+  it('fetchLastProductionReadyAt reads the newest READY production deployment', async () => {
+    const { calls, fetchImpl } = captureFetch(
+      fakeResponse(true, 200, {
+        deployments: [{ ready: 1_700_000_000_000, createdAt: 1_699_999_000_000 }],
+      }),
+    )
+
+    await expect(
+      fetchLastProductionReadyAt({
+        token: 'token',
+        projectId: 'prj_test',
+        teamId: 'team_test',
+        fetchImpl,
+      }),
+    ).resolves.toBe(1_700_000_000_000)
+
+    expect(calls[0]!.url).toContain('/v6/deployments?')
+    expect(calls[0]!.url).toContain('projectId=prj_test')
+    expect(calls[0]!.url).toContain('teamId=team_test')
+    expect(calls[0]!.url).toContain('target=production')
+    expect(calls[0]!.url).toContain('state=READY')
+  })
+
+  it('evaluateDeployCooldown surfaces API failures', async () => {
+    const { fetchImpl } = captureFetch(
+      fakeResponse(false, 403, { error: { message: 'forbidden' } }),
+    )
+
+    await expect(
+      evaluateDeployCooldown({
+        token: 'token',
+        projectId: 'prj_test',
+        teamId: 'team_test',
+        fetchImpl,
+      }),
+    ).rejects.toThrow(/403/)
+  })
+})