1. User Input Received

-> A user submits text (e.g., a chat message, prompt, or document).

  1. Apply Protection Rules

-> The system checks the input against predefined rules:

  • Toxicity, Tone, Sexism, Prompt Injection, Data Privacy
  1. Run Input Through Protect

-> The Protect Engine evaluates the input using these rules.

-> If a rule is triggered, the system flags it.

  1. Determine Outcome

PASS: If all checks pass → The input is approved and processed as normal.

FAIL: If any check fails → The input is blocked and a failure message is generated.

  1. Return Results

-> The system provides a detailed response, including:

  • Status (Passed/Failed)
  • Checked Rules (Which rules were evaluated)
  • Failed Rule (If applicable)
  • Reason (Explanation of failure)
  • Processing Time
  1. Deploy in Production

-> Protect is integrated into the system to filter inputs in real time.

How toOverview