Zero-Noise Change Detection Explained
What Is Zero-Noise Detection?
Traditional website monitoring compares raw HTML. This means every A/B test, analytics script, or dynamic CSRF token triggers a false alert. DiffFlow's Zero-Noise engine filters out all of that.
How It Works
1. Fetch — Our SafeFetcher downloads the page through SSRF-validated channels
2. Normalize — We strip <script>, <style>, hidden inputs, and dynamic attributes (csrf, nonce, token, session, timestamp)
3. Selector — Apply your CSS selector to extract only the content you care about
4. Hash — Compute a SHA-256 hash of the normalized content
5. Compare — If the hash changed from the last check, you get an alert
What Gets Filtered
- All
<script>tags — JavaScript changes don't matter - All
<style>tags — CSS changes are visual, not content - Hidden form inputs — CSRF tokens and nonces change every request
- Dynamic attributes —
data-session,data-ts,nonce,csrfvalues - Whitespace normalization — extra spaces don't count as changes
CSS Selector Precision
Without a CSS selector, DiffFlow monitors the entire visible text. With a selector like .price-value, you watch exactly the price. With .stock-status, you watch just the inventory.
Text Filtering
Need to only see alerts when "In Stock" appears? Use a text filter. Combined with CSS selectors, text filters let you build surgical monitoring rules.
The Result
Our users report 95% fewer false alerts compared to generic monitoring tools. When a DiffFlow alert fires, you know real content changed.