MCP Audit Gateway

This is a small helper that sits between your AI assistant and the other tools it uses, and writes down every action in a signed log. If you ever wonder what your AI actually did, or you need to prove it later, this gives you a record you can check. It is handy for anyone who lets an AI assistant touch real files, code, or accounts and wants a paper trail.

What this MCP server does

You set it up once by putting a short prefix in front of the command that starts your existing MCP server. From then on, whenever your AI asks another tool to do something, that request passes through this helper first. The helper lets the request go through unchanged and notes down the tool name, the arguments, how long it took, and whether it worked. Each note is signed and linked to the previous one, so if anyone edits or removes a line, the check fails. You can watch the log live or verify the whole chain later.

Click to zoom

What you can do with it

• See a live stream of every tool call your AI makes
• Check that the log has not been edited or shortened
• Look up what arguments were passed to a specific tool call
• See which calls succeeded and which failed
• Add access rules and rate limits for a team setup
• Send the audit data to your monitoring tools
• Wrap any existing MCP server without changing how it works

What it gives back to you

You get a list of tool calls with the time, the tool name, how long it took, and whether it worked. Each line has a short code you can use to look it up. When you run the verify command, you get a simple report: how many records were checked, how many were valid, and whether anything looks off. In the full gateway mode you also get access decisions and traces sent to your monitoring tools.

Before you start

Who it's for

Anyone who lets an AI assistant take real actions on their files, code, or accounts and wants a record they can check later, especially small teams and people in security or compliance roles.