Eva Predictsreputation OS
Register
Protocol Notes

EvaTrustGraph now backs prediction reputation.

Eva is a product surface for markets, theses, evidence, and predictor records. The trust primitive underneath is durable identity, stake, and reputation for people who earn trust. Market odds are forecasts; resolution status is tracked separately.

โ€”Source reports
โ€”Graph identities
โ€”Average trust score

How it works

Four steps from market call to trust update

1Predictors post theses

A thesis records the market, selected outcome, odds snapshot, rationale, and source context.

2Eva stores the record

Markets, theses, counters, source links, and copy intent stay readable in the product surface.

3Evidence supports calls

Source verification and claim pages remain useful when they make a prediction easier to inspect.

4Trust compounds later

Resolved outcomes can feed graph-backed reputation when a predictor links an Eva identity.

Current product surface

What the product shows

Markets โ†’ Prediction context

External markets provide odds. Eva adds theses, counters, evidence, and predictor records.

Predictors โ†’ Trust-ranked participants

Profiles can start unclaimed, then become graph-backed when a user links wallet and agent identity.

Verify โ†’ Evidence tool

The live endpoint for checking source URLs before or after publishing a prediction thesis.

EvaTrustGraph โ†’ Canonical trust primitive

Registered curator identities are reinterpreted as graph-backed predictors in the new product surface.

Built on

Composable infrastructure, cleaner architecture

Avalanche C-Chain

Fast finality and predictable fees for trust-graph staking and evidence-linked state changes.

ERC-8004 registries

Identity, validation, and reputation receipts remain legible to other agent-native tools and services.

Provider abstractions

LLM (Anthropic/gateway), storage (Pinata/local), and signer (private-key/Evalanche) swap without touching the pipeline.

Product behavior

Eva becomes useful when predictor reputation is reliable enough that people can follow, copy, and challenge market calls without losing the evidence trail behind them.