Membership tier
Controls what features are available and whether enforcement is binding.
Provider agility
Choose which engine evaluates actions for safety verdicts (built-in, external, or custom).
Org-wide controls
Confirm tier status and whether enforcement features are enabled for the org.
Membership tiers
The dashboard exposes three organization tiers. The selected tier determines whether verdicts are purely logged, advisory, or binding.
| Tier | What it does | Enforcement behavior |
|---|---|---|
| Audit | Basic on-chain logging without AI evaluation. | Manual — no system-issued verdicts. |
| Audit+ | Adds AI advisory verdicts without active enforcement (no blocking). | Advisory only — decisions are recorded but do not gate execution. |
| Enforcement | Full protection including auto-blocking, the Approval Queue, and red line enforcement. | Binding — BLOCK refuses execution; HOLD routes to operator review. |
AI security provider
Atbash supports provider agility: you can choose which engine evaluates actions for safety verdicts.
Atbash (built-in)
The native, zero-config engine. Choose this when you want a managed default without external keys.
External providers
Options for Google Security AI, Microsoft Defender AI, and OpenAI. These typically require an API key and (for some providers) an endpoint.
Custom endpoint
Point the system to a proprietary security model by providing a custom URL and API key. This is useful when you operate your own OpenAI-compatible inference endpoint or internal risk engine.
Organization info
The bottom of the page includes read-only metadata that helps operators confirm what configuration is in effect.
Read-only metadata
- Org name
- Tier status (Audit / Audit+ / Enforcement)
- Whether enforcement is globally enabled for the org