ohernandezdev/jevmod on GitHub
- Stars
- 0
- Forks
- 1
- Language
- Python
- License
- MIT
- Last push
- 23 Sep 2026
Read from GitHub on . Counts change daily.
How it uses Jev
resp = self.client.system_one(state=state, questions=questions)
self.requests += 1
self.input_tokens += getattr(getattr(resp, "usage", None), "input_tokens", 0) or 0
self.judged_messages += len(to_judge)
# Offset by one: the real messages start at m1 because m0 holds the lead filler.
for i, (m, text) in enumerate(to_judge, start=1):
scores = {c: _p(resp.answers[f"{c}_{i}"]) for c in cats}
custom = {name: _p(resp.answers[f"custom__{name}_{i}"]) for name in custom_rules}
self.cache[_key(text, m.channel_topic, cats, custom_rules, m.context)] = (now, scores, custom)
out[m.id] = Verdict(m.id, scores, True, "jev", custom)
elif to_judge:
for m, _ in to_judge:
out[m.id] = Verdict(m.id, {}, False, "no categories enabled")
return [out[m.id] for m in messages]Use casesLLM guardrails
