Complete AI Training
Sign inBecome a member

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

MCP server · Security

Derbent MCP server

by tunahanaliozturk

One gate that checks, approves and records every tool call your coding agents make.

Derbent is a small helper program that sits between your coding agents (like Claude Code, Codex, Copilot CLI and Antigravity CLI) and the tools they want to use. It decides whether each tool call is allowed, denied, or needs your approval first, and it writes everything down. It is handy for anyone who lets an AI agent edit files or run commands on their computer and wants a say in what happens.

What is an MCP server? The 30-second version

On its own, your AI can only chat. An MCP server is a small helper program that gives your AI a new skill or a connection to something outside the chat. Derbent is that helper here: your coding agents connect to it, and it acts as one gate for all their tool calls, whether those are MCP tools or the agent's own built-in tools like the shell. You set the rules, Derbent checks each call against them, and it keeps a record you can verify later.

What this MCP server does

You start your coding agent as usual, and it connects to Derbent. When the agent wants to do something, like read a file or run a command, that call goes to Derbent first. Derbent looks at your rules and decides: allow it, deny it, or ask you. If it asks, the call waits in a terminal window until you press a key to approve it or it times out. Every call, allowed or not, gets written into a tamper-evident log called receipts.

What you can do with it

  • Set rules that allow, deny or ask about specific tools and commands
  • Approve or deny waiting tool calls from a terminal window
  • See a live list of what your agents are doing
  • Check the receipt log to see if anything was changed
  • Put your other MCP servers behind the same gate
  • Let agents leave notes and tasks for each other per project
  • Stop an agent stuck in a loop with a call budget

Try asking your AI

  • “Show me the tool calls waiting for my approval”
  • “Explain how Derbent would decide a git push command from Claude”
  • “List the receipts from the last hour for the codex agent”
  • “Verify that the receipt chain has not been changed”

What it gives back to you

You get answers in your chat or terminal, like a list of waiting calls, a summary of recent receipts, or a plain yes or no on whether the log is intact. When Derbent explains a decision, it shows each rule and why it matched, ending with the verdict. Approved or denied calls show up in the terminal UI as they happen.

Before you start

What you need

  • The Derbent binary for your system, downloaded from the releases page or installed with Homebrew or Scoop
  • One or more supported coding agents, like Claude Code or Copilot CLI
  • A terminal to run the setup and the approval window

Good to know

Derbent is not a sandbox: an agent that can already run shell commands as you can get around it, so approvals guard against mistakes and prompt injection but not against an agent that can run derbent approve itself.

Install it with your AI

Add Derbent MCP server to your AI, no technical skills needed

You don't install anything by hand. You copy one prompt, paste it into an AI that can work on your computer, and it checks, installs and connects the server for you, asking you when it needs something.

Sign in to get the install prompt

Members get a ready-made prompt that lets the Claude desktop app check Derbent MCP server, install it and connect it for them, step by step. You don't need any technical skills: you copy, paste and answer a few questions. Your connected AI can also find and install any of the 4,609 MCP servers here for you.

Sign in Become a member

Who it's for

Developers and anyone who lets AI coding agents run commands or edit files on their machine and wants control and a record of what they do.

Jobs it suits