Skip to main content

Simulation run on this device. Scenario values are not sent anywhere — the engine is fully local.

prompting studio · free

AI Input Privacy Scrubber

Pattern-scan inputs for keys, PII, and customer data before they leave.

Did that prompt just leak a secret?

Engine 1.0.0 · studio
Findings
4
High severity
2
Characters saved
12
  • high
    OpenAI-style key × 1

    sk-abc1234567890XYZ

  • high
    AWS access key × 1

    AKIAIOSFODNN7EXAMPLE

  • medium
  • medium
    Phone (intl) × 1

    +1 415 555 0134

Redacted output — safe to paste into a prompt:

Contact ***@***.*** or +***-***-****. Use key sk-***REDACTED*** and AKIA***REDACTED***.

Redact before the text enters the context window; once sent, a secret is already leaked.

Method

  • Pattern-scans for API keys, tokens, emails, phones, SSNs, IPs, and private-key blocks, then emits a redacted copy.
  • Redaction is a last line of defense: the only safe secret is one that never entered the prompt.