Blockchain2023
COMPOSABLESUBGRAPHS
Composable subgraph architecture for reusable DeFi protocol building blocks.
2
Source DEX subgraphs
1
Composed GraphQL endpoint
4+
Shared pool entity types
SCROLL
Composable subgraph architecture for reusable DeFi protocol building blocks.
Uniswap v3 and PancakeSwap v3 emit overlapping pool lifecycle events, yet institutional consumers wanted a single GraphQL surface for TVL, volume, and fees — not two schemas that drift every upgrade.
Two thin source subgraphs (per-DEX factories, NFPM, tuned start blocks) feed a composed layer that merges entities, rolls protocol-level counters, and exposes one endpoint for cross-venue analytics.
Compose `kind: subgraph` sources via IPFS CIDs, graft Pancake state for faster sync, and normalize Mint/Burn/Swap/Collect into shared Protocol / Pool / Token entities without double-counting liquidity.
Isolate each DEX’s factory, NFPM, and event surface with minimal cross-talk.
Wire `kind: subgraph` sources with pinned IPFS CIDs and explicit schema imports.
Graft Pancake state where applicable to shorten cold-sync time on new networks.
Map Mint/Burn/Swap/Collect into shared Protocol/Pool/Token with consistent decimals.
Protocol-level counters for combined TVL and fee revenue without double counting NFPM positions.
Uniswap v3 and PancakeSwap v3 unified in one composed schema so dashboards could answer “aggregate DEX volume” without client-side joins across two subgraph deployments.
Uniswap v3 and PancakeSwap v3 unified in one composed schema so dashboards could answer “aggregate DEX volume” without client-side joins across two subgraph deployments.
Composed sources let each DEX evolve independently while the umbrella schema stayed stable.
Grafting turned multi-day Pancake syncs into hours on new chains without rewriting mappings.
Immutable subgraph artifacts made promotions reproducible and audit-friendly.
Shared entity IDs required strict factory ordering rules to avoid hash collisions in merged pools.
We take on a small number of projects at a time. If the problem is hard, we're interested.