From 512205ed882177de4e1e04988cb903eb4e0bdf8f Mon Sep 17 00:00:00 2001
From: ydd039 <ydd039@sohu.com>
Date: Fri, 15 May 2026 14:09:01 +0800
Subject: [PATCH] add: beginner tutorial - Protect Your AI Agent

---
 docs/tutorials/protect-your-ai-agent.md | 98 +++++++++++++++++++++++++
 1 file changed, 98 insertions(+)
 create mode 100644 docs/tutorials/protect-your-ai-agent.md

diff --git a/docs/tutorials/protect-your-ai-agent.md b/docs/tutorials/protect-your-ai-agent.md
new file mode 100644
index 0000000..a12329d
--- /dev/null
+++ b/docs/tutorials/protect-your-ai-agent.md
@@ -0,0 +1,98 @@
+---
+title: "Protect Your AI Agent in 5 Minutes: A Beginner's Guide to APort"
+published: false
+description: "Learn how to add safety guardrails to your AI agents with APort - an open-source authorization infrastructure. No experience needed!"
+tags: [ai, security, opensource, tutorial]
+---
+
+# Protect Your AI Agent in 5 Minutes
+
+AI agents are powerful, but they can also be dangerous. Without proper guardrails, an AI agent might delete files, make unauthorized API calls, or access sensitive data. Enter **APort** - an open-source authorization infrastructure that keeps your agents in check.
+
+## What is APort?
+
+APort is like a security guard for your AI agents. Before your agent executes any action, APort checks if that action is allowed. If it is, the agent proceeds. If not, the action is blocked.
+
+The core concepts are simple:
+
+- **Passport**: Your agent's identity card - defines what it is and what limits it has
+- **Policy**: Rules that specify what actions are allowed
+- **Verify**: The checkpoint where APort checks an action against policy
+
+## Quick Start
+
+Let me show you how to add APort to a Python agent in under 5 minutes.
+
+### Step 1: Install APort
+
+```bash
+pip install aport-agent-guardrails
+```
+
+Or for Node.js:
+
+```bash
+npm install @aporthq/aport-agent-guardrails
+```
+
+### Step 2: Add APort middleware
+
+For a FastAPI application, it's just a few lines:
+
+```python
+from fastapi import FastAPI
+from aport_middleware import APortMiddleware
+
+app = FastAPI()
+app.add_middleware(APortMiddleware)
+```
+
+### Step 3: Define your first policy
+
+Create a simple policy that prevents your agent from running dangerous commands:
+
+```python
+policy = {
+    "name": "safe-commands",
+    "rules": [
+        {"action": "exec.run", "allow": True, "if": "command not in ['rm -rf', 'drop table', 'shutdown']"},
+        {"action": "file.write", "allow": True, "if": "path.startswith('/tmp/')"},
+        {"action": "network.request", "allow": False, "unless": "domain in WHITELIST"}
+    ]
+}
+```
+
+### Step 4: Verify actions
+
+```python
+result = await aport.verify(
+    passport_id="passport_abc123",
+    action="exec.run",
+    resource={"command": "ls -la"}
+)
+
+if result["allow"]:
+    # Safe to execute!
+    print(f"Action allowed: {result}")
+else:
+    print(f"Blocked: {result['reasons']}")
+```
+
+## Why This Matters
+
+As AI agents become more autonomous, the risk of unintended actions grows. APort gives you:
+
+1. **Accountability**: Every action is logged and verified
+2. **Safety**: Dangerous actions are blocked automatically
+3. **Auditability**: Full trail of what your agent tried to do
+4. **Flexibility**: Policies can be updated without changing code
+
+## Next Steps
+
+- Try the [Go example](https://github.com/aporthq/aport-integrations/tree/main/examples/hello-world/go)
+- Check out [APort docs](https://docs.aport.io)
+- Join the community on [GitHub](https://github.com/aporthq/aport-integrations)
+
+---
+
+*APort is open-source (Apache 2.0) and ready for production use. Start protecting your agents today!*