How scores work
Reputation scoring is part of the Stellar expansion. The product intends to evaluate activity, contribution, and ecosystem relationships. The current registry stores the resulting numbers; the production calculation service and exact scoring model are not implemented in these repositories.
Social reputation
The intended social analysis considers public activity and the relationship graph around an X profile. This includes who follows the account, whom the account follows, and the relevance of those relationships to Stellar and the broader web3 ecosystem.
A useful evaluation needs context and evidence quality. The product brief does not define numerical weights, a trusted-account list, time windows, normalization, or an anti-manipulation formula. These should not be inferred from example profile values.
GitHub reputation
The intended developer analysis considers work performed, project participation, and contributions relevant to Stellar. It can distinguish contribution history from a simple activity count.
The repositories do not define a scoring formula for commits, pull requests, reviews, issues, repositories, or ecosystem relevance. Listing these as potential evidence does not mean an implemented service currently scores each of them.
Onchain reputation
The intended onchain analysis uses activity associated with the connected Stellar account. No production indexer, transaction classifier, anti-sybil model, or published scoring weights are present here.
An account balance or token holding alone is not a published reputation rule.
Stored values
| Field | Stored type | Allowed range |
|---|---|---|
social | u32 | 0 to 4,294,967,295 |
github | u32 | 0 to 4,294,967,295 |
onchain | u32 | 0 to 4,294,967,295 |
| Total | Sum in u64 | 0 to 12,884,901,885 |
These are storage limits, not a product scale such as 0 to 100 or 0 to 1,000. The total is an unweighted sum. A missing record is distinct from a registered record containing zeros.
Who controls updates
Only the registry admin can replace scores. An update replaces all three fields together and immediately affects future checks. The contract verifies the writer's authority; it does not verify the underlying social or contribution evidence.
Current score records do not contain a model version, evidence hash, verification flag, calculation time, or expiry time. A consuming application that needs freshness or evidence must obtain it through a separately defined system.
The scoring service will need published policies for refreshes, identity changes, disputed evidence, provider failures, and manipulation. Those policies are open product work, not current guarantees.