
Masquerade
par postralai
Privacy firewall for LLMs that redacts sensitive data from PDFs before they are sent to AI models.
What it does
Masquerade acts as a privacy layer between your sensitive documents and Large Language Models. It allows users to process PDFs by identifying and redacting sensitive information (names, emails, entities) locally or via an isolated AI platform, ensuring that raw private data never reaches the LLM provider.
Tools
redact_pdf: Processes a PDF file path, detects sensitive entities, and creates a redacted version of the document.preview_redactions: Provides a summary of masked data and redaction counts per page for user verification.
Installation
Add the following to your claude_desktop_config.json:
{
"mcpServers": {
"pdf-redaction": {
"command": "python",
"args": ["/path/to/mcp_pdf_redaction.py"],
"env": {
"TINFOIL_API_KEY": "your_api_key"
}
}
}
}
Supported hosts
- Claude Desktop
Installation rapide
curl -O https://raw.githubusercontent.com/postralai/masquerade/main/setup.sh && bash setup.shInformations
- Tarification
- free
- Publié
- 4/19/2026
- étoiles






