A threat model identifies an untrusted parser handling privileged requests. What should be prioritized in the design?

1. Minimize parser privilege and validate boundary inputs
2. Hide its error messages as the only defense
3. Increase server capacity before defining security boundaries
4. Run the parser with domain-wide administration