AI Terminal

┌─────────────────────────────────────────┐
│  username@server:~$                     │
│                                         │
│  [Your command output appears here]     │
│                                         │
│  Ask AI: "How do I check disk space?"   │
└─────────────────────────────────────────┘

You: My website is slow, what should I check?

AI: Let me help you diagnose performance issues. I'll check:
1. CPU usage
2. Memory consumption
3. Disk I/O
4. Running processes

Suggested command:
$ top -bn1 | head -20

Would you like me to run this? [Run] [Explain more]

$ find /var/log -name "*.log" -mtime +7 -delete

// System prompt structure
const systemPrompt = `
You are a Linux system administration expert helping via SSH terminal.
Current server: ${hostname}
User: ${username}
Working directory: ${pwd}

Rules:
1. Prefer safe, read-only commands for initial diagnosis
2. Explain commands before suggesting destructive operations
3. Use precise syntax for the detected shell (${shell})
4. Consider the user's apparent skill level
5. Suggest alternatives when possible

Respond with either:
- A command to execute (prefixed with $)
- An explanation or answer in natural language
`;

interface AiContext {
  // Server information
  hostname: string;
  os: "ubuntu" | "centos" | "debian" | "amazon-linux";
  architecture: "x64" | "arm64";

  // Session state
  currentDirectory: string;
  recentCommands: string[];
  environmentVariables: Record<string, string>;

  // User preferences
  expertiseLevel: "beginner" | "intermediate" | "expert";
  preferredShell: "bash" | "zsh" | "fish";
}

type RiskLevel = "SAFE" | "CAUTION" | "DESTRUCTIVE";

const riskIndicators = {
  SAFE: ["ls", "cat", "ps", "top", "df", "du", "grep", "echo"],
  CAUTION: ["systemctl restart", "kill", "pkill", "chmod"],
  DESTRUCTIVE: [
    "rm -rf",
    "dd",
    "mkfs",
    "fdisk",
    "parted",
    "DROP DATABASE",
    "DELETE FROM",
    "TRUNCATE",
  ],
};

function classifyRisk(command: string): RiskLevel {
  // Implementation checks against risk indicators
  // Returns appropriate risk level
}

You: nginx failed to start

AI: Let me check what's wrong with nginx.

[Running]: systemctl status nginx

● nginx.service - A high performance web server
   Loaded: loaded (/lib/systemd/system/nginx.service; enabled)
   Active: failed (Result: exit-code)

Error: Port 80 is already in use!

Suggested fix:
1. Check what's using port 80
2. Stop the conflicting service
3. Restart nginx

[Check port 80] [Show me how to fix]

You: Install PostgreSQL and create a database

AI: I'll help you install PostgreSQL and set up a database.

Step 1: Install PostgreSQL
$ sudo apt update && sudo apt install postgresql postgresql-contrib

Step 2: Start the service
$ sudo systemctl start postgresql

Step 3: Create database
$ sudo -u postgres createdb myapp

Step 4: Create user
$ sudo -u postgres createuser -P myuser

Would you like me to execute these steps? [Execute all] [Step by step]

You: Find all 404 errors in nginx logs from today

AI: I'll search for 404 errors in today's nginx access logs.

Command:
$ grep "$(date +%d/%b/%Y)" /var/log/nginx/access.log | grep " 404 "

Results:
192.168.1.100 - - [13/Apr/2026:10:15:32 +0000] "GET /missing-page.html HTTP/1.1" 404 564
192.168.1.101 - - [13/Apr/2026:11:22:15 +0000] "GET /old-link.php HTTP/1.1" 404 564

Summary: Found 2 requests resulting in 404 errors.

You: Delete old log files

AI: I'll help you clean up old logs. Let me show what will be deleted:

[Dry Run Results]:
Would delete: /var/log/app/old.log (2.3 MB, 45 days old)
Would delete: /var/log/nginx/access.log.2.gz (156 MB, 60 days old)
Total space to free: 158.3 MB

[Execute] [Cancel] [Show me the command first]