MCP server · Security
pod MCP server
Records what your AI agent really does, writes a tight permission policy from it, and checks every action against it.
pod is a helper that watches what your AI agent actually does for a few days, then writes a small permission policy based on those real actions. After that, it sits between your agent and its tools and decides what is allowed. It is handy if you use AI agents that can read files, run commands or change things, and you want to keep them on a short leash without writing rules by hand.
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 an app or service. This one connects your AI to pod, a permission and audit tool that sits in front of your agent's tools. When your AI wants to do something, pod checks it against your policy first, so the AI can act only within the limits you set.
What this MCP server does
You run your agent normally for a day or two while pod records the tool calls it makes. Then you ask pod to compile a policy from that record, and it shows you a before and after list of what got tightened, added, removed or loosened. When you switch pod into enforcement mode, every tool call from your agent passes through it first, and pod answers allow, approve or deny. High-risk calls pause and wait for a human to approve them. Every call is written into a tamper-evident log, so you can later check that nothing was quietly changed.
Click to zoomWhat you can do with it
- Record what your AI agent actually does for a few days
- Compile a least-privilege policy from those real tool calls
- See a clear diff of what got tightened, added, removed or loosened
- Enforce the policy at the gateway, with deny beating approve beating allow
- Pause risky calls and approve them by hand from a second terminal
- Verify the audit log to check nothing was edited
- Export a portable evidence bundle you can hand to someone else
Try asking your AI
- “Scan my machine and show me which MCP servers are exposed”
- “Draft a least-privilege policy from the calls you recorded and show me the diff”
- “Verify the audit chain and tell me if anything was changed”
- “Export an evidence bundle I can send to my client”
What it gives back to you
You get back plain answers in the chat: lists of servers and tools, a policy draft with a diff table, counts of what got tightened or removed, and pass or fail results from the audit check. When a call needs approval, pod tells you what is waiting and lets you approve or deny it. The evidence export gives you a folder of files with a one-page report and a hash manifest.
Before you start
What you need
- Node.js 22.13 or newer and git
- pnpm to build from source
- An AI agent or MCP client you want to watch over
Good to know
pod can change your agent's configuration files when you use the takeover step, so read the before and after it shows you, and know that it keeps backups you can revert.
Install it with your AI
Add pod 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 pod 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 4,066 MCP servers here for you.
Who it's for
People who run AI agents that touch files, run commands or change things, and want a small, honest permission policy plus a record of what happened.
