LINKScan

Discovery Of Hot Contracts By Category & Blockchain
Description of the smart contract lifecycle and the data pipeline
Dapp based graph analysis to understand and map smart contracts
Interactive data analysis on every Chainlink customer on chain
Dapp & Token metadata accessible directly in the explorer

Inspiration/Motivation

Blockchain data is available and transparent for everyone to see and verify. It is also the primary source used to make good decisions as protocol developers. Unfortunately, this data is still hard to understand, not used enough and its access fragmented. I wanted to build a blockchain explorers that works for multiple teams. The vision was to build a minimalistic, simple, and efficient Discovery & Deep Dive dashboard to help more teammates get "their hand dirty" with data. That is why LINKScan is a light ⁽¹⁾ internal blockchain explorer build by the Data Analytics team with these 4 key principles in mind:

Accessibility: Provide a no-code experience for querying, monitoring, and auditing on-chain data for everyone.
Oracle Focus: Focus on Chainlink products and services relalated analytics.
Self-Service: Provide quick self-service capabilities for analysts for efficient analytics.
Data Quality: Enable deep dive on internal models and datasets to perform debugging, quality check, and data labelling.

LINKScan achieves the above by being a blockchain explorer that is:

DApp & Hybrid-Smart Contract Centric: While many explorers exist out there, and are used by many internally and externally, they focus on the flow of transactions and blocks. LINKScan puts +5000 dApps and their +380K ⁽²⁾ smart contracts at the center of the exploration. This is because dApps are the starting point of any anyltics deep dive within Chainlink Labs teams (SAs, Finance, Product, Marketing, and Go-To-Market).
Visual Exploration: LINKScan can analyze on-chain data like dApp deployment patterns through pivot tables, self-service plotting, and graph analysis thus making it accessible for anyone to deeply understand what is happening cross-chain. LINKScan is also very easy and friendly to use. Think of it as if Google Sheets + Etherscan/Poygonscan/etc + Chainalysis had a baby...
Multi-Chain & Cross-Chain Exploration: Blockchain explorers are still stuck in a mono-chain world (each blockchain has its own explorer: ehterscan.io for Ethereum mainnet, Snowtrace.io for Avalanche etc). This makes for a fragmented "deep dive flow" for data citizens as Chainlink is deployed on multiple chains and care for cross-chain users. LINKScan aggregates the data by smart-contract first and, second, provides a breakdown by all available blockchains a contract is on ⁽³⁾.
Automated & ML-Powered: LINKScan automates the querying and scrapping of multiple third-party data source ⁽⁴⁾. The team is building ML driven SC idenfication models (clustering) to map +10K DApps to +5M smart contracts on-chain ⁽⁵⁾. ML SC classification models will be used to increase the categories & use case coverage for each smart contract. Data Analysts can also speed up their work by leveraging LINKScan SQL generation capabilities ⁽⁶⁾ and preview log extracts of dApp contracts.

Features:

With LINKScan you can:

Discovery

Discover the most active dApps by blockchain
Discover the most active Chainlink Users By Product by blockchain in the last period ⁽⁷⁾
Filter by number of on-chain transactions or dApp users ⁽⁸⁾
Filter by 18 categoriese dApps category ⁽⁹⁾

DApp Explorer

Dapp Search

Dive Deeper into a sepcific dApp deployed contracts:
- Search Any dApp in free form text
- Filtering dApp results by data source

Token Overview

Checking metdata information from Coingecko, CoinmarketCap, and others on all tokens deployed by a dApp
Navigating directly to resources online (Github, community, Website, etc)

Blockchain & Oracle Activity

Building interactively pivot tables & charts about on all the contracts deployed by a dApp
Building interactively pivot tables & charts about on all the Chianlink activity by a Chainlink User

Contract Explorer

Studying specific SC by going through their code
Generating SQL queries & preview of contracts logs

Graph Explorer

Vizualize the graph of dApp deployers accross multiple chains.

Future Work:

Add support for more chains
Add support for compiled smart contracts
Extend category classification to more categories
Build internal pipelines for dApps users and TVL metrics
Extend contracts & wallets labeling to unknown entities (i,e idetify "DeFi App", "NFT trader", etc)
Incorporate SC Identification into pivot table and graph analysis analysis automatically
Add more parameters to analyze Product Usage (use cases classification, data feeds (pair) breakdown, date rangee)
Add collaboration to leverage the explorer for data correction and labelling
Deploy a JupyterHub instance and setup security settings to open LINKScan to more internal teams

Notes:

(1): LINKScan is built in Python in ~1000 code lines as a Jupyter Notebook. It leverages the data pipeline and results of SCAI (Smart Contract AI) a machine learning project kickstarted by the ML team to create the most accurate map of hybrid-SC corss-chain.
(2): Only active (have had at least 1 transaction on chain) and (source code) verified (by blockexplorers like Etherscan) contracts has been loaded for now. Team is gathering verified contracts from Polygon & Binance next & will expand to take into account internal transactions as well as decompiled/non-verified contracts.
(3): Ethereum, Avalanche, Optimism, and Heco. The explorers can load other chains if the filtering described in (2) is not applied.
(4): Data has been gathered/scrapped from Coingecko, Coinmarketcap, Defillama, DappRadar, and StateofDapps. Internal datasets has been also created to fill in the gap. Etherscan/Bscscan public labels are next on the team's list.
(5): +5M have been mapped to +8K sperate entities (sperate "deployers') already, of which +1M has been mapped to known dApps. Because of filtering decribed in (2) only +250K contracts are loaded in this instance.
(6): SQL Generation is done by mapping all contracts ABIS code interfaces) to dataset indexed daily by the DA team from multiple chains.
(7): Only Price Feeds Users from 2022-05-01 2022-05-23
(8): number of users from DappRadar a of starting of May
(9): DeFi, Synth, Games, 'Exchanges', 'Collectibles', 'NFT', 'Gambling', 'Staking', 'Marketplaces', 'Collectibles & NFTs', 'Yield Farming', 'Yield Aggregator', 'Governance','Platform','Farming', 'Layer 2', 'Play to Earn', 'Metaverse'