26 May 2026
FILE № W-0042-DOCS
📚 Case file documentation

The Casebook.

Technical documentation, architecture, and API reference for Walytics — Walytics Holmes Edition.

📖 What is this

Overview

Walytics is a real-time analytics and exploration platform for Walrus decentralized storage on Sui. It indexes blob storage activity, visualizes patterns, and provides AI-powered insights via Gemini. Built with Tatum's Sui RPC for on-chain data access and the Walrus protocol for storage.

🏗️ How it's built

Architecture

Frontend
Next.js 16 (App Router), Custom CSS, Recharts, Framer Motion
Backend API
Next.js API Routes (serverless, Node.js runtime)
Blob Data
Blockberry API (same source as walruscan.com)
RPC / Health
Tatum Sui RPC (health check + fallback indexer)
AI Engine
Google Gemini (gemini-2.5-flash-lite)
Storage
Walrus Protocol (snapshot write-back)
Deploy
Vercel
1User → Next.js App → API Routes → Blockberry API → live Walrus blob data
2API Routes → Tatum Sui RPC → health check + fallback indexing
3API Routes → Gemini AI → chat responses & analytics reports
4API Routes → Walrus Publisher → snapshot blobs stored on-chain
🔌 Routes

API Endpoints

GET /api/analytics

Aggregated Walrus metrics (blobs, publishers, trends)

GET /api/blobs

Paginated list of stored blobs

POST /api/chat

AI chat + report generation (Gemini)

POST /api/tatum

Generic proxy to Tatum Sui RPC

POST /api/snapshot

Save analytics snapshot to Walrus (write-back)

🔗 On-chain data

Tatum Integration

  • Health check: sui_getLatestCheckpointSequenceNumber — live indicator on dashboard (polls every 15s)
  • RPC proxy: /api/tatum endpoint with method whitelist + per-IP rate limiting
  • Fallback indexer: suix_queryEvents for BlobRegistered when Blockberry is unavailable
  • Endpoint: https://sui-mainnet.gateway.tatum.io (x-api-key auth)
🦭 Decentralized storage

Walrus Integration

  • Read: Live blob data via Blockberry API (same indexer as walruscan.com)
  • Write: Analytics snapshots stored as Walrus blobs — immutable, verifiable on walruscan.com
  • Dogfooding: Walytics analytics are archived on the very network they analyze
  • Erasure code: RS2 (RedStuff — Walrus default)
🤖 Holmes intelligence

AI Features

  • Model: gemini-2.5-flash-lite (free tier)
  • Chat: Natural language queries about live Walrus storage data
  • Reports: One-click analytics summary — archived as a Walrus blob on completion
  • Context: Real-time metrics injected into every query (Blockberry live data)
  • Character: Walytics Holmes — detective persona with Walruscan.com citation rules
⚙️ Get running

Setup

1Get Tatum API key at https://dashboard.tatum.io
2Get Gemini API key at https://aistudio.google.com
3Get Blockberry API key at https://blockberry.one (for live blob data)
4Clone repo: git clone https://github.com/EdCryptoFi/Walytics
5Copy .env.local.example → .env.local and fill keys
6npm install && npm run dev
GitHub Repo →Live Demo →Tutorial →