Reduce Accidental Data Leaks to AI Tools
Prompt Censor helps detect and warn about sensitive data before you send prompts to ChatGPT, Claude, or other AI tools. All checks run locally in your browser.
Important: Prompt Censor uses pattern-based detection and cannot guarantee 100% coverage. Always review your content before submitting sensitive information.
Protected Data Types
Pattern-based checks designed to help you catch common high-risk content. Results may include false positives or misses depending on format and context.
Passwords
Helps flag common password patterns and risky credential-like inputs before they are sent to AI tools.
API Keys
Helps identify API keys, access tokens, and authentication credentials in real-time using local pattern matching.
Tokens
Helps detect OAuth tokens, JWTs, and other token-like strings. Coverage depends on format and context.
SSH Commands
Helps prevent accidental sharing of SSH keys, commands, and server credentials typed into AI chats.
SQL Credentials
Helps flag database connection strings, usernames, and other potentially sensitive DB-related content.
Other Sensitive Data
Supports additional patterns (for example: cards, IDs) and custom rules you configure. False positives and misses may occur.
How It Works
Three simple steps to reduce accidental leaks
You Paste Content
Type or paste your prompt into ChatGPT, Claude, or another supported AI website.
Local Check
Prompt Censor scans the text locally using rule-based pattern matching to highlight risky strings.
You Decide
If something looks sensitive, you can edit, remove, or stop before sending. You stay in control.
Local processing only
Prompt Censor does not send your input to our servers. Settings are stored locally in your browser.
Detection is best-effort and may not catch every sensitive pattern in all contexts.
Built on Trust & Security
Privacy-focused architecture, without promising perfect detection
Local-only Analysis
Your input is analyzed locally in the browser to show warnings before you submit content to a third-party site.
Fast by Design
Lightweight pattern checks run instantly without server calls. Performance may vary by page and device.
Open Source
Transparent codebase you can review. Community feedback helps improve rules over time.
No Server Collection
We don't send your input to our servers. Settings are stored locally using browser storage.
Install Prompt Censor
Available for Chrome, Edge, and Brave. Always review your content before sending prompts to third-party AI services.