ShadowDiff - Dual-traffic shadow testing & SQL diffing for modern backend

by•
Thousands of engineering teams are using AI to modernize legacy monoliths into cloud-native Go and TypeScript services. But the silent 5% of invisible edge cases (serialization asymmetries, float precision drift, missing SQL audit writes, and ORM N+1 query loops) causes CTOs to reject production cutovers. ShadowDiff solves this with zero-latency Envoy traffic mirroring, mathematical 3-way Diffy noise cancellation, database side-effect diffing, and automated regression test synthesis with LLMs.

Add a comment

Replies

Best
Maker
📌

Hey Product Hunt community! 👋

I'm Vishal, creator of ShadowDiff.

Over the past year, many engineering teams have turned to AI coding models to modernize legacy backends (Java Spring monoliths, Python 2 Django) to Go.

While AI refactors syntax with ~95% accuracy, the remaining 5% of hidden edge cases (JSON serialization quirks, float precision drift, missing SQL audit table writes, and ORM N+1 queries) stop CTOs from cutting over live traffic.

Traditional traffic shadowing fails because non-deterministic fields (timestamps, UUIDs) create 99% false alarm noise.

We built ShadowDiff in Go to give teams mathematical certainty:

⚡ Envoy line-rate packet mirroring (zero latency impact)

🎯 Mathematical 3-way Diffy noise cancellation (P1 ⊕ P2 filter)

💾 SQL mutation & ORM N+1 query loop detection

🛡️ Ephemeral Copy-on-Write DB sandboxing (zero DB pollution)

🔒 Automated PII redaction (Luhn-validated credit cards, SSNs, JWTs)

🤖 Automated regression test synthesis with local Ollama/vLLM & Gemini

🌐 Live Cloud Sandbox:

📦 Open Source:

I'd love to hear your thoughts and feedback!