FRCEM Mentor

The two tools form a clean generate/grade pair with no overlap: get_practice_question produces a stem with hidden answer, check_answer grades a submitted answer. An agent can never reasonably confuse one for the other.

Both names follow the same verb_noun snake_case pattern (get_practice_question, check_answer) with domain-appropriate verbs (get for retrieval, check for validation). No mixed conventions.

Two tools is thin in absolute terms, though each earns its place in the quiz loop. The surface is minimal enough that agents may want adjacent helpers (topic listing, session reset) that don't exist.

The core lifecycle — serve a question, accept an answer, return answer plus explanation and guideline references — is fully covered, and seen_ids supports avoiding repeats. Gaps are minor: no topic/curriculum enumeration, no progress or score tracking across a session.