math-identity-check

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.30). The math-identity-check skill ingests outsider-provided lhs and rhs free-text expressions directly in verify_identity(...) → _build_side(lhs/rhs) → _parse_sympy(expr) (and optionally compile_formula(expr)), so attacker-controlled text is parsed and evaluated at runtime.