ChainSentry

Demo

About

The problem it solves Normally people interested in doing detailed on-chain analytics would have to: - Subscribe prominent on-chain forensics tools like Chain-Analysis, nansen which are : cost prohibitive, slow to adapt (still have a subset of EVM / non EVM chains and wallets supported) and only depend on the - Blockchain data warehousing and analytics tools like dune, covalent and then define their custom queries to analyse the data: but still they need to write complex queries and you can run recursive queries (like finding the common addresses across the transaction addresses that have 2 way transactions and so on….). Thus we provide a swiss knife like tool for users who can explain their intent and start from one address , and then iteratively apply the selection of strategies (based on the addresses , NFT’s and other ). Why agents and how? : Because after the creation of every branch of transaction, the corresponding transaction logs (both assets or contract calls) and type of address (proxy , EOA, smart account) create enormous search space of possibilities of the analysis . thats why it was important to create the memory component: cache that stores incrementally the address , address category and then gives user more details in the metadata so that they can inferr the causal relations of the activities of address and take corresponding decisions. User Interaction and Data Flow ## Overview Our application enables users to investigate blockchain addresses, visualize transaction flows, and analyze on-chain activities across multiple networks using blockchain data scraping, graph visualization, and AI-powered analysis. ## User Journey - **Authentication**: Self-verification or wallet-based authentication - **Main Flow**: Investigation setup → address entry → network selection → analysis → visualization → reporting ## Core Workflows ### Address Investigation Users enter addresses and select networks to fetch transaction data. The system displays visualizations and address details, allowing users to select nodes/transactions and apply analysis strategies. ### Strategy Execution Analysis strategies run sequentially: from idle → ready → active → completed, with each strategy updating the visualization upon completion. ### Transaction Analysis Different transaction types (token transfers, NFT transfers, bridge transactions, contract interactions) follow specific analysis paths to identify counterparties and assess risks. ## Components - Strategy Control Panel - Graph Visualization - Details View - Address List - AI-assisted Chat Box ## Technical Architecture Built on Next.js with API routes connecting to browser scraping services and AI analysis tools. ## Use Cases ### Tracking Suspicious Funds Trace fund movements and highlight suspicious patterns using the "Follow the Money" strategy. ### Cross-Chain Analysis Track addresses across multiple blockchains by identifying bridge transactions and creating comprehensive cross-chain visualizations. ### Smart Contract Audit Analyze contract details, interactions, and connected addresses to generate risk assessments and audit reports. The application provides an intuitive interface for blockchain investigations, enabling users to gain insights into transaction patterns, identify suspicious activities, and track fund flows across blockchain networks. The project architecture and development process # Blockchain Forensics Solution ## Overview Our blockchain forensics platform helps investigators track and analyze transactions across multiple networks using browser automation, graph visualization, and AI analysis. We address cross-chain visibility, pattern recognition, address categorization, and investigation efficiency challenges. ## Core Functionality ### 1. Blockchain Data Scraping - Multi-chain support (Ethereum, Bitcoin, Polygon, BSC) - Automated extraction of transaction history and balances - Entity recognition for exchanges, mixers, and DeFi protocols ### 2. Graph Visualization - Interactive network graphs with force-directed layouts - Color-coded nodes based on address types - Transaction flow visualization with temporal analysis ### 3. Analysis Strategies - Follow the Money: Trace fund flows from source to destination - Cluster Analysis: Identify related addresses through patterns - Cross-Chain Tracking: Track funds across different blockchains ### 4. AI-Powered Insights - Transaction summarization in human-readable format - Anomaly detection for suspicious activities - Risk scoring based on transaction history ## Implementation Highlights - Browser automation with Stagehand/Browserbase - Structured data model using TypeScript and Zod - WebGL-based graph rendering for large transaction networks - LangChain integration with GPT-4o for intelligent analysis ## Tech Stack - Frontend: Next.js, React, Tailwind CSS - Data Processing: TypeScript, Zod - AI: LangChain, OpenAI GPT-4o - Authentication: Self-verification and wallet-based Our solution enables investigators to efficiently track funds, analyze patterns, and generate insights across blockchain networks with a modular architecture that adapts to the evolving landscape. Product Integrations # Blockchain Forensics: Product Integrations ## Blockchain Data Services ### Browserbase/Stagehand - **Role**: Browser automation for scraping blockchain explorers - **Value**: Extracts transaction data from any explorer without direct API access ### Blockchain Explorers - **Services**: extensive use of crosschain - **Value**: Provides transaction history, address details, and token transfers ### LiFi API - **Role**: Cross-chain transaction tracking - **Value**: Follows funds across different blockchain networks ## AI & Analysis Services ### OpenAI GPT-4o - **Role**: Natural language processing and pattern analysis - **Value**: Powers transaction summarization and anomaly detection ### Fireworks AI (Claude Models) - **Role**: Alternative AI provider for specialized analysis - **Value**: Complements GPT models with different analytical strengths ### LangChain - **Role**: AI orchestration framework - **Value**: Simplifies complex AI workflows and tool integration ## Authentication & Security ### Privy - **Role**: Authentication and wallet connection - **Value**: Secure, wallet-based authentication with minimal friction ### Self Protocol - **Role**: Self-verification of users - **Value**: KYC-like verification without storing sensitive data ## Data Storage & Management ### Vercel Postgres - **Role**: Structured data storage - **Value**: Stores user data and investigation results ### Redis - **Role**: Caching and real-time data - **Value**: Improves performance and enables collaboration Key differentiators and uniqueness of the project We have significant differentiations from the Trade-offs and shortcuts while building ## Key Tradeoffs: ### Decision of using browser automation of access the address's crosschain interactions : 1. We initially had thought of using the primitive blockchain data accessing services (RPC's like alchemy , dRPC ) and others but had to shelve the idea due to the constraints on the throughput and the normalization logic. 2. We had developed the integration with browserbase in order to parse the transaction logs to showcase the crosschain bridges , but its not performant and for browserbase there needs to be checks in order to scrape data --> check its authenticity --> and then render all of them. also for concurrent scraping we need to pay hefty fees . thus for the demo we just did the run on the pre-existing scrapped address.