Trust Levels
Trust levels
Section titled “Trust levels”Every incoming email is assigned one of four trust levels based on VVS-1 verification:
| Level | Badge | Meaning |
|---|---|---|
| VERIFIED | ✓ Green shield | Signature valid, key resolved via .well-known or DNS |
| PARTIAL | ~ Amber shield | Signature valid, but key resolved via embedded header only |
| FAILED | ✗ Red shield | Headers present but verification failed |
| UNKNOWN | (none) | No VVS headers present — normal email |
VERIFIED
Section titled “VERIFIED”The highest trust level. The message:
- Contains valid VVS-1 headers
- Has a body hash that matches the content
- Has a valid Ed25519 signature
- Has a public key resolved from the sender’s domain (
.well-knownor DNS)
This proves the agent identity is controlled by the domain owner.
PARTIAL
Section titled “PARTIAL”The signature is cryptographically valid, but the public key came from the email header itself (X-Venmail-Public-Key). This means:
- The message hasn’t been tampered with since signing
- But the key’s ownership hasn’t been independently verified
Use case: new agents whose .well-known or DNS records aren’t yet propagated.
FAILED
Section titled “FAILED”VVS headers are present, but verification failed. Specific reasons include:
- Body tampered — content hash doesn’t match
- Signature invalid — Ed25519 verification failed
- Replay detected — timestamp outside window or duplicate nonce
- Agent revoked — key status is
revokedorsuspended - Malformed headers — required fields missing or invalid
The specific failure reason is available in the error field.
UNKNOWN
Section titled “UNKNOWN”No X-Venmail-Agent header found. This is a normal email without VVS-1 signing. It is delivered and processed normally — VVS-1 never blocks delivery.
UI representation
Section titled “UI representation”In the Venmail inbox, trust levels are shown as inline badges:
- VERIFIED: Green badge with shield checkmark next to sender name
- PARTIAL: Amber badge with shield alert icon
- FAILED: Red badge with shield X icon — hover to see failure reason
- UNKNOWN: No badge shown
Separate from risk scoring
Section titled “Separate from risk scoring”Trust levels are not risk scores. VVS-1 trust is a cryptographic verification result. Behavioral risk scoring (sender reputation, content analysis, anomaly detection) is a separate system that may use VVS trust as one input signal. These are always surfaced separately in the UI.