A risk number is only as trustworthy as the data underneath it. This page traces that data end to end — from the on-chain source, through the pipeline that normalizes it, to the API response that hands it back with its provenance attached. The differentiator is not that we have the data; it is that the heavy normalization work is already done, and every answer can tell you where it came from.Documentation Index
Fetch the complete documentation index at: https://docs.alterscope.org/llms.txt
Use this file to discover all available pages before exploring further.
The pipeline
Alterscope sources directly from the chain and the protocols — never from a second-hand aggregator’s API — then normalizes once so you don’t have to.Protocol connectors
Dedicated connectors read each supported protocol’s on-chain state and events: lending markets, vaults, curator governance, pool reserves, and oracle configuration. Each connector understands one protocol’s contracts, so the values it returns are the protocol’s own, not a re-derived approximation.
Ethereum archive node
EVM reads run against a full Ethereum archive node, which retains historical state at every block. This is what makes block-level history possible — a price, peg, or liquidity value as of a specific past block, not just the latest snapshot. That backfilled history is what the advanced risk models need to be meaningful.
Time-series store
Oracle feeds and price history land in a time-series database. Each feed is registered with its expected update cadence, and every observed price is stored with its timestamp — this is the substrate for staleness detection and for the peg-history and oracle pages.
Knowledge graph
Assets, protocols, vaults, oracles, and wallets are modeled as a knowledge graph of their real relationships. Network analysis over this graph produces the contagion, concentration, and capital-flow signals that a per-asset view cannot see. See Graph intelligence.
Governance & event indexers
Governance actions — MetaMorpho curator events such as timelock, guardian, allocator, cap, fee, owner, and queue changes — are indexed as they happen, and oracle adapters are continuously classified for staleness behavior. These feed the vault-governance timeline and the oracle classification score.
”Normalized” is the product
Raw on-chain data is heterogeneous: every protocol encodes positions, decimals, and oracle wiring differently. Alterscope does that reconciliation once — resolving decimals, mapping each market to its assets and oracle, classifying feed behavior, and rolling positions up into portfolio terms — so the number you receive is already comparable across protocols and chains. That pre-processing is the work you would otherwise repeat for every integration.Provenance is a first-class field
You do not have to take it on faith that the pipeline ran. Every wrapped response can name the sources behind its own numbers, in the agentic metadata block:meta._agentic.sourceslists the data sources that backed the response — for example["timescaledb", "archive_node"].meta._agentic.field_confidence.<field>carries a per-fieldmethodandsources, so a single number can declare exactly how it was derived and from where.meta._agentic.freshnessreports when the data was computed, how old it is, and its plain-language status.
Where to go next
- How fresh, and is it fit to act on? Freshness & quality.
- What’s covered, and where are the gaps? Coverage & gaps and the Coverage matrix.
- How the numbers are computed from this data: Methodology overview.