Building Agentic Commerce #3: Reading Trust Score v4.1

Trust Score v4.1 combines three weighted dimensions with an optional external-assurance bonus and confidence-tier caps. Each response identifies the engine used; some responses fall back to legacy v2. A legacy v2 component breakdown does not explain an overall score produced by v4.1.

Read engines.overall and engines.breakdown before interpreting a score. overall uses 0–1; trust_score_v41 uses 0–100 when available. With engines.mixed=true, the breakdown does not explain overall. insufficient_data is missing usable evidence, not a measured zero; the public route can return a degraded v2 fallback.

An agent that consumes Trusteed scores can use them as one input to a decision. Other agents may use reviews, visual information or other sources. A score does not authorize a purchase, replace consent or guarantee fulfillment.

Weights depend on confidence and are renormalized across applicable dimensions. The external-assurance bonus is applied after the base score, before the tier cap. For the published base model, caps are 65 for cold_start, 85 for human_established, 92 for agent_verified_thin and 100 for agent_verified. Read the live methodology for criteria and any enabled extensions; a tier name is not an identity-verification certificate.

Read verified, measured, proxy, self_declared, not_applicable and synthetic_blocked as distinct evidence qualities. Missing data must not be presented as a measured zero. If the v4.1 overlay is disabled, fails or produces no value, the public route can serve legacy v2. A degraded v2 result is not comparable to v4.1 as if both had the same evidence requirements.

GET https://api.trusteed.xyz/api/v1/stores/{slug}/trust-score returns the public score. Resolve the merchant and its data source from /.well-known/merchant-index.json; do not assume the same path on another host reads the same database. Verify the response signature using the issuer’s trusted JWKS. A valid signature authenticates the response, not the quality of all source data.

Observed on 2026-09-13 for demo-store, a sandbox store; this is a dated example, not its current score: ```json { "overall": 0.447, "trust_score_v41": 44.7, "confidence_level": "cold_start", "engines": { "overall": "v4.1", "breakdown": "v2", "mixed": true } } ```

Search eligibility, result ranking and checkout enforcement are separate decisions. Consult the contract of the specific endpoint and merchant configuration. A public score can come from a different engine than a stored value used by search. Labels such as VERIFIED are score bands, not proof of identity or permission to skip confirmation.

Merchant Trust Score describes the merchant. Buyer-agent and seller-agent trust concern other actors and their own controls. Do not reuse a merchant score as the agent’s authorization or copy thresholds between these policies.

The legacy v2 model is a weighted sum on a 0–1 scale, scheduled every 6 hours. Its components may still appear beside a v4.1 overall when engines.mixed=true. Only use this breakdown to explain a number when engines.overall identifies v2. The methodology endpoint lists the weights and labels this scope legacy_v2.