diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
new file mode 100644
index 0000000..2543160
--- /dev/null
+++ b/.github/workflows/ci.yml
@@ -0,0 +1,20 @@
+name: CI
+
+on:
+  push:
+    branches: ["**"]
+  pull_request:
+
+jobs:
+  validate:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-node@v4
+        with:
+          node-version: 20
+          cache: npm
+      - run: npm ci
+      - run: npm run typecheck
+      - run: npm run build
+      - run: npm test
diff --git a/README.md b/README.md
index 0553821..3d96edb 100644
--- a/README.md
+++ b/README.md
@@ -88,6 +88,24 @@ Props: `listening` · `canTranscribe` · `waveform` · `elapsed` · `onStart` ·
 
 ---
 
+
+## Production Readiness
+
+VoiceDraft is publishable as a package today, but production apps should ship with a few guardrails:
+
+- **Capability fallback:** Always check `canTranscribe` and render a typed-input fallback path.
+- **Permission handling:** Pass `onError` to show user-friendly prompts when microphone permission is denied.
+- **Post-processing:** Treat transcript text as a draft and run your own validation/moderation before submit.
+- **Cross-browser QA:** Verify behavior in Chrome, Edge, and Safari where Web Speech support differs.
+
+### Launch checklist for AI apps
+
+1. Add product copy and onboarding around browser mic permission.
+2. Persist drafts in your app state so transcripts are recoverable.
+3. Add server-side logging for speech failures and cancel/confirm rates.
+4. Add your preferred cloud STT fallback for unsupported browsers.
+5. Run CI (`npm run typecheck`, `npm run build`, `npm test`) on every PR.
+
 ## Browser Support
 
 Requires `SpeechRecognition` or `webkitSpeechRecognition`. Use `canTranscribe` to gracefully disable the mic path when unavailable.
diff --git a/package.json b/package.json
index 5e551f3..00f9828 100644
--- a/package.json
+++ b/package.json
@@ -37,7 +37,8 @@
     "build": "npm run clean && tsc -p tsconfig.build.json && node scripts/copy-styles.mjs",
     "prepack": "npm run build",
     "typecheck": "tsc --noEmit",
-    "benchmark:stt": "node scripts/benchmark-stt.mjs"
+    "benchmark:stt": "node scripts/benchmark-stt.mjs",
+    "test": "npm run build && node tests/format.test.mjs"
   },
   "keywords": [
     "voice",
diff --git a/tests/format.test.mjs b/tests/format.test.mjs
new file mode 100644
index 0000000..e1903bd
--- /dev/null
+++ b/tests/format.test.mjs
@@ -0,0 +1,10 @@
+import assert from "node:assert/strict";
+import { formatElapsed } from "../dist/utils/format.js";
+
+assert.equal(formatElapsed(0), "0:00");
+assert.equal(formatElapsed(65), "1:05");
+assert.equal(formatElapsed(600), "10:00");
+assert.equal(formatElapsed(-1), "0:00");
+assert.equal(formatElapsed(Number.NaN), "0:00");
+
+console.log("format tests passed");