Profile & policy

Profile (edit): empty, complete onboarding first

#Rule (first match wins)Strategy
1unsafe (safe < 0.5) → warn, no LLMunsafe
2off-topic (on_topic < 0.5) → warn, no LLMoff_topic
3teacher role → lesson-planning helpteacher
4wants answer + no attempt → ask for their attempt, no LLMshow_work
5check my work → verify onlyverify
6frustrated → worked exampleworked_example
7wants answer + attempt shown → one hinthint
8wants to understand → explainexplain
9anything else → brief replychat
You: this app sucks

safe yes (0.87) · on-topic no (0.15) · intent other (1.00) · attempt no (0.02) · frustration frustrated (0.70)
Rule: off-topic (on_topic < 0.5) → warn, no LLM

⚠️ That seems off-topic for general academic and learning topics. Try asking something about general academic and learning topics.
Jev 302 ms 837 in · 113 out tok · $0.000035 · no LLM call · total 302 ms
Jev raw output · 302 ms
{
  "output": {
    "safe": 0.87,
    "on_topic": 0.15,
    "intent": "other",
    "attempt": 0.02,
    "frustration": 2
  },
  "confidence": {
    "intent": 1.0,
    "frustration": 0.7
  },
  "probabilities": {
    "intent": {
      "understand": 0.0,
      "get_answer": 0.0,
      "check_work": 0.0,
      "other": 1.0
    },
    "frustration": {
      "0": 0.0,
      "1": 0.0,
      "2": 0.7,
      "3": 0.3
    }
  },
  "scores": {
    "frustration": 2.3
  }
}
Thinking…
Session usage · $0.000035
ModelRequestsInput tokOutput tokCostAvg / requestPer 1k requests
Jev 1 837 113 $0.000035 950 tok $0.0352