Blockhead 2026

Demo

About

In early 2026, after five years iterating on the same codebase over dozens of hackathons, I decided to dive headfirst into the world of agentic software factory engineering and rebuild Blockhead from scratch. The result: a local-first interface with a highly accurate, irreducible, internally consistent ontology capable of modeling integrations with any decentralized protocol, blockchain ecosystem, wallet, node or data source. Browse artifacts across the decentralized web, connect existing wallets across any ecosystem, interact with protocols of all shapes and sizes, and plug and play with your own nodes and data providers leveraging Blockhead's extensive library of adapters.

How it's made

Blockhead's codebase models every stack layer explicitly and modularly. Provider sources handle external requests and data queries. Resolvers map responses into native entities. Selectors uniquely identify entities and route URLs with distinct sets of fields. Local-first collections cache data and feed the UI views. For ETHOnline 2026, I extended its native data model with Messari Standardized Subgraphs and adding Ledger support to its wallet interface. The Graph integration queries live Messari Standardized Subgraphs. Protocols, pools, block observations, and ordered pool assets use native selectors. Historical requests preserve the deployment and block hash. Reported schema versions remain separate from the versions expected by our deployment configuration. Missing data is an error, not an empty result. Ledger uses Device Management Kit, Ethereum Signer Kit, and the official Speculos transport. The wallet adapter checks signing authority and verifies message signatures. Browser tests exercise device approval, rejection, and persisted request history. In the demo, I have an automated harness drive a browser to demonstrate data queries and local-first client side caching. It also drives an instance of Ledger's Speculos emulator, clicking through screens to connect to Blockhead's UI and facilitate a signing request. Uniswap V3 and Sushi V3 financial observations use the same view and query pattern. Each observation retains its exact values, source deployment, schema version and block. Reloading an observation requests that block, rather than silently replacing it with current data. A separate Uniswap view interprets pool state, including its raw token ratio, fee configuration and oracle capacity. The Ledger flow reviews a message on the official Speculos emulator, signs it and saves a request that can be reopened after reload. Once I have the stack layers, data modeling and automation harnesses fully nailed down, I plan to continue building out the UI layer and implement immersive graph visualizations on top of the client-side cache.