API & Command Permission Checks — decisions, not generations

Agent frameworks increasingly need a fast, deterministic gate before executing a tool: "may this action run on this state?" Decision models answer with a calibrated probability you can threshold — no prompt-hacking a chat model into yes/no, no output-token cost, no parsing failures.

Why decision models

• Deterministic gate in front of every tool call
• Policy as criteria — auditable, versioned in code
• No output tokens, no response parsing
• Works with the official typesafe-sdk

Keep your policy in the question criteria, keep the state as structured JSON, and branch directly on the typed answer. Every check is billed by input tokens only.

Runnable example

curl https://api.easyjev.online/v1/systemone \
  -H "Authorization: Bearer sk-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "typesafe/jev-latest",
    "state": {"user": {"plan": "pro"}},
    "questions": {
      "is_premium": {"type": "noul", "criteria": {"true": "…", "false": "…"}}
    }
  }'