Aegis Privacy MCP
Run agent-bound text and document flows through a PII and secrets sanitization MCP before prompts, logs, or OCR pipelines leak customer data.
Overview
Aegis-Privacy-MCP (zero-leak) is an MCP server for the Ship phase that sanitizes PII and secrets—with OCR-aware merging—before agent workflows leak sensitive data.
What is this MCP server?
- Aegis-Privacy-MCP (zero-leak) gateway for PII and secrets sanitization
- Advanced OCR merging logic for mixed text extracted from images or scans
- Remote SSE MCP with optional EIP-3009 premium payment-signature header
- Designed as a sanitization layer autonomous agents can call before downstream tools
- Catalog entry io.github.Evozim/zero-leak at version 1.0.0
- Version 1.0.0 per server.json-style manifest
- 1 hosted remote SSE MCP endpoint
- 1 documented premium secret header: payment-signature (EIP-3009)
What problem does it solve?
Agents ingest prompts, tickets, and OCR’d docs that accidentally carry emails, tokens, and credentials into logs or external models.
Who is it for?
Builders shipping agent features that touch user uploads, support transcripts, or scanned documents who need a consistent sanitization step.
Skip if: Organizations that require on-prem only data planes with no remote MCP, or teams that need certified compliance attestation without their own policy layer.
What do I get? / Deliverables
After you add the MCP gateway, agents can redact or block sensitive fields upstream so safer text proceeds to analysis, storage, or reply generation.
- Central MCP sanitization step for agent pipelines
- Reduced accidental PII/secrets exposure in prompts and OCR merges
- Documented remote endpoint and header requirements for production agents
Recommended MCP Servers
Journey fit
How it compares
Privacy sanitization MCP gateway, not a generic secret scanner CLI or penetration-test skill pack.
Common Questions / FAQ
Who is zero-leak (Aegis Privacy MCP) for?
Solo builders and small teams running MCP agents on customer-facing or internal data who must strip PII and secrets before LLM calls or logging.
When should I use zero-leak?
Use it before shipping agent tools that read OCR output, pasted logs, or ticket text, and when hardening operate workflows that reuse production snippets.
How do I add zero-leak to my agent?
Add the remote SSE MCP server URL (https://zero-leak-mcp.vercel.app/api/mcp) in your client configuration and configure the payment-signature header if your plan uses EIP-3009 premium access.