Contexto

Problem Statement

An observability layer for Hedera agents that reads wallet footprints across the chain, filters spam wallets, and enriches context for better interactions. Each wallet leaves signals — age, transaction history, token diversity, dApp usage, and activity patterns. Our system converts these footprints into profiles with scores like spam likelihood, engagement, loyalty, and whale value. Agents use these insights to block junk, focus on real users, and personalize responses. Developers gain dashboards to view wallet clusters, risk scores, and behavior heatmaps. The result: safer, smarter, and more profitable agent-wallet interactions powered by transparency, trust, and actionable intelligence.

Solution

Contexto is built using the Hedera AI Agent SDK as the foundation for the agent layer. Since the AI agent only accepts signatures from Hedera’s native wallet (HashPack), users joining the platform are first onboarded by creating or connecting a HashPack wallet.Once the AI side is set up, the wallet analysis pipeline begins. Using The Graph’s Token API, Contexto collects the full footprint of a wallet across EVM-compatible networks in order to rank and evaluate it. This helps distinguish trustworthy wallets from spam-associated ones. For example, a healthy wallet might show consistent swaps, liquidity provision, or long-term holdings, while a spam wallet might show repetitive transfers or interactions with suspicious contracts.After evaluation, the results are passed through a summarizer module that interprets the raw transaction data and produces clear conclusions about the wallet. These insights are then sent back to the AI agent so it can adjust its behavior dynamically. Finally, Hypergraph is used to store the details of the agent interactions in a structured way, serving as a database layer for context persistence.