Your personal operating system for career transition.

Build a structured prompt template with system/user role separation

Prompt injection attacks happen when user-controlled content reaches the system role. The first defense is strict structural separation.

What you are building

Implement a PromptBuilder class that:

1. Accepts a system_template (a string with {variable} placeholders for trusted application data)
2. Accepts template_vars (a dict of trusted values to fill in)
3. Accepts user_input (raw user text, always kept in the user role — never interpolated into system)
4. Returns a messages list [{"role": "system", "content": ...}, {"role": "user", "content": ...}]
5. Raises ValueError if user_input text appears inside the rendered system content

Requirements

• Template variable substitution should use Python's str.format_map
• If a required template variable is missing, raise KeyError with a clear message
• The user input must never appear in the system message
• Strip leading/trailing whitespace from both rendered messages

Example usage

builder = PromptBuilder(
    system_template="You are a support agent for {company}. Tone: {tone}.",
    template_vars={"company": "Acme Corp", "tone": "professional"},
)
messages = builder.build(user_input="What is your refund policy?")
# messages == [
#   {"role": "system", "content": "You are a support agent for Acme Corp. Tone: professional."},
#   {"role": "user", "content": "What is your refund policy?"},
# ]