Trust Score¶
Every governed agent in Trust3 carries a Trust Score — a numeric rating on a 1.0–10.0 scale that summarizes how well that agent meets your organization's governance requirements. It gives IT, compliance, and risk teams a shared signal for prioritization without requiring them to read individual policy results for every asset.
Score display¶
Trust Score is shown as a value out of 10 (for example 4.6 / 10 on the asset detail sidebar). Higher is better. Each agent also has a trust level band derived from the published score, and a risk level label derived from that trust level for prioritization.
| Trust level | Risk level | Published score | What it means |
|---|---|---|---|
| High | Limited | 7.75 and above | Agent is well-governed. No immediate action needed. |
| Medium | Low | 5.95 – below 7.75 | Gaps present. The owner should review. |
| Low | Medium | 4.15 – below 5.95 | Significant gaps. Remediation is needed. |
| Critical | High | Below 4.15 | Severe governance gaps or multiple failing signals. Prioritize remediation. |
Risk level is the inverse framing of trust level: stronger governance posture maps to lower operational risk. Trust3 derives it automatically (critical trust → High risk, low trust → Medium risk, medium trust → Low risk, high trust → Limited risk) and exposes it on inventory and asset-detail APIs as riskLevel.
In addition to the band, each agent carries a trust status:
| Status | What it means |
|---|---|
| Trusted | Agent is governed and meets active policy requirements |
| Unverified | Agent has been discovered but not yet fully evaluated |
| Untrusted | Agent has critical violations or is ungoverned |
Where you see it¶
- Inventory table — each row shows the trust score, trust level badge, and derived risk level, color-coded: green (High / Limited risk), amber (Medium / Low risk), orange (Low / Medium risk), red (Critical / High risk)
- Asset detail panel — sidebar shows the score as X.X / 10 with the trust level badge; the Overview tab Risk Status card reflects the derived risk level
- Dashboard — stats strip shows counts of critical and low-trust agents across your estate
- GIA — ask "show me low-trust agents" or "which agents are untrusted?" for natural-language answers grounded in the live inventory; see GIA
What moves a score¶
Score drops when:
- A policy violation is opened and unresolved
- The agent loses an assigned owner
- The agent has no description or scope documentation
- An unapproved model is detected
Score recovers when:
- Violations are resolved and closed
- Ownership is assigned or confirmed
- Missing fields are completed
- Policy evaluation passes on the next discovery cycle
Related documentation¶
- Policies — how violations connect to evaluation and trust score
- Core Concepts — Policy and remediation lifecycle