privacyfence/privacyfence

Overview

Open-source privacy and approval gateway that sits between AI assistants and business systems using the Model Context Protocol. It adds human approval, local PII checks, and audit logging for integrations like Gmail, Drive, Slack, Salesforce, and Jira. Written in Python.

Best for

Best for
Teams that want a self-hosted approval and PII gate for AI assistants on SaaS tools.

Use cases

• Require human approval before an AI assistant sends emails or edits Drive files
• Run local PII checks on AI-generated requests before they reach SaaS APIs
• Keep an audit trail of AI assistant actions across connected business tools

How to use

Tested with

Claude Desktop, Claude Code, ChatGPT

Notes

Open-source privacy and approval gateway that sits between AI assistants and business systems using the Model Context Protocol. It adds human approval, local PII checks, and audit logging for integrations like Gmail, Drive, Slack, Salesforce, and Jira. Written in Python.

1 stars on GitHub. Last updated 2026-10-08. Licensed Apache-2.0.

Use cases

• Require human approval before an AI assistant sends emails or edits Drive files
• Run local PII checks on AI-generated requests before they reach SaaS APIs
• Keep an audit trail of AI assistant actions across connected business tools

Pros

• Open source and self-hosted, so data stays under your control
• Covers common business systems including Gmail, Drive, Slack, Salesforce, and Jira
• Adds a human-in-the-loop layer to MCP-based AI workflows

Cons

• Very early stage with minimal community adoption (1 star)
• Requires self-hosting and manual setup
• Only works with assistants that support MCP

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.