← All Tools
LLM Pre-flight Redactor
Scan a prompt before you paste it into ChatGPT, Claude, Gemini, or Copilot. Catch PII, API keys, internal hostnames, and other things that should never end up in a third-party model.
100% client-side. The text you paste here never leaves your browser. No telemetry, no storage. The whole point is that it stays private even from us.
⚠ Heuristic, not a guarantee. Pattern matching catches obvious leaks (typed credentials, common PII formats, AWS-style keys). It cannot catch everything — not paraphrased PII, not novel secret formats, not data that describes something sensitive without quoting it. Always review the redacted output yourself before sending.
What to scan for
Custom keywords (one per line)
Project codenames, customer names, internal product names. Will be replaced wholesale.
Why pre-flight matters
- Vendor data-handling clauses do not protect you from accidental disclosure of regulated personal data, trade secrets, or credentials.
- Logs and stack traces often carry session IDs, IPs, file paths, and schema details you don't want in someone else's training corpus.
- Customer names and internal codenames are surprisingly common in pasted "fix this for me" prompts.
- Many enterprises now run a copilot allow-list / DLP layer on egress. This tool gives the same safety net for your draft prompts.
- Aligns with NIS2 Art. 21, DORA Art. 9, GDPR Art. 32 (technical and organisational measures).