AiNIR Trust Gate is the unified decision surface for the pre-v1 public demo.
It does not integrate AiNIR with any external compiler or runtime. It consolidates the existing AiNIR gates into one decision artifact so a reviewer can answer a simple question:
Can this AI-generated semantic draft move toward lowering, receipt replay, or future handoff?
The Trust Gate summarizes these checks:
- Strict Draft AST
- Safety Registry Resolution
- Evidence Ledger Binding
- Operation Spec Binding
- Capability Contract
- Workflow Semantic Profile
- Trusted Execution Context
- Transaction Binding
- Policy Core
- Lowering Eligibility
| Status | Meaning |
|---|---|
passed |
The draft passed the bounded public-demo gates. Lowering may be allowed if all lowering checks also pass. |
refused |
The draft is structurally valid, but one or more safety-critical gates refused it. |
hold |
The draft needs clarification, evidence, or review before it can move forward. |
invalid |
The draft did not satisfy the strict intake format. |
{
"status": "refused",
"executable": false,
"lowering_allowed": false,
"handoff_allowed": false,
"failed_gates": [
"evidence_ledger",
"capability_contract"
],
"reasons": [
{
"rule_id": "EVIDENCE_SELF_ATTESTED",
"severity": "critical",
"message": "Verified claims require ledger-bound evidence."
}
]
}python -m ainir trust-gate examples/create_user_outbox_safe/draft.yaml --json --out-dir /tmp/ainir_trust_gateUnsafe or malformed drafts produce status: refused, status: hold, or status: invalid and must not be lowered.
handoff_allowed is a generic future extension signal. It is not tied to any specific downstream consumer in the current pre-v1 public demo.