← Directory / Safety gateways
sint-ai/sint-protocol
Runtime governance for AI agents with approval tiers and physical constraints.
Score75%
Unknown6/12
TypeOpen source
Last checked2026-10-16
12-point checklist
| # | Criterion | CISA | Score | Evidence |
|---|---|---|---|---|
| C1 | Actuation capability 2 pts if: Cannot write/actuate, or writes are clearly isolated | P2, P4 | mediates | |
| C2 | Read-only by default 2 pts if: Write/method tools off unless explicitly enabled | P4 | Not documented publicly / not yet verified | |
| C3 | Write guardrails 2 pts if: Per-tag allowlist plus value/rate limits or interlock checks | P4 | physical constraint enforcement | |
| C4 | Human-in-the-loop approval 2 pts if: Critical writes need explicit human confirmation | P4 | T0-T3 approval tiers | |
| C5 | Authentication 2 pts if: Authenticated sessions to the OT endpoint; no anonymous default | P3, P4 | api_key/authn | |
| C6 | Encryption & cert verification 2 pts if: Encrypted transport with server certificate verification | P4 | TLS mentioned | |
| C7 | Audit logging 2 pts if: Every agent call/write logged durably | P3, P4 | audit receipts | |
| C8 | Network placement 2 pts if: On-prem/edge/air-gapped or push/DMZ flow; no inbound IT→OT path | P2, P4 | Not documented publicly / not yet verified | |
| C9 | Data leaving the site 2 pts if: No telemetry; plant data can stay on-prem | P2 | Not documented publicly / not yet verified | |
| C10 | Fail-safe & safety independence 2 pts if: Fails closed; safety systems stay independent | P4 | Not documented publicly / not yet verified | |
| C11 | Vendor security disclosure 2 pts if: Published security policy / PSIRT | P3 | Not documented publicly / not yet verified | |
| C12 | Supply-chain hygiene 2 pts if: Versioned/signed releases; pinned installs | P3 | Not documented publicly / not yet verified |
Sources