Building ShadowPaste in public: Ask me anything about AI security and MCP

Hey everyone 👋

I'm building ShadowPaste, a local-first zero-trust MCP gateway that helps developers use AI coding agents on real repositories without exposing secrets.

ShadowPaste focuses on:

• Secret virtualization

• AI-safe workspaces

• Zero-trust MCP gateway

• Dangerous tool-call blocking

• Audit trails and risk scoring

• Credential restore after AI editing

I'm launching soon and would love feedback from developers.

Questions I'd love to discuss:

- How do you currently protect API keys in Claude Code or Cursor?

- Have you ever accidentally leaked secrets to an AI tool?

- Which security features are missing in today's AI coding workflow?

Ask me anything and share your thoughts 🚀