Reversing Obfuscated Javascript logo

Reversing Obfuscated Javascript

Community
trilwu
reversing-obfuscated-javascript

Reverse engineer minified, bundled, and obfuscated browser/Node JavaScript — unpacking webpack chunks, recovering source from sourcemaps, undoing obfuscator.io string-array and control-flow obfuscation with webcrack/synchrony/restringer, and locating a signing or crypto routine in a live bundle via Chrome DevTools. Use when a page ships a huge minified bundle, when code is full of _0x hex identifiers and a rotated string array, when a .js.map is reachable, or when you must find where a request signature or token is computed.

Overview

Publishertrilwu
Repositorysecskills
Skill namereversing-obfuscated-javascript
Stars
144
Forks
15
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 trilwu on GitHub. Read the source before you install it.

Installation

Install the Reversing Obfuscated Javascript 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/trilwu/secskills.git /tmp/secskills
mkdir -p .claude/skills
cp -r /tmp/secskills/secskills-core/skills/reversing-obfuscated-javascript .claude/skills/reversing-obfuscated-javascript
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Reversing Obfuscated Javascript 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 Reversing Obfuscated Javascript 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 Reversing Obfuscated Javascript 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.

Reversing Obfuscated JavaScript

Client-side JavaScript ships the whole program to the reader — there is no stripped binary here, only code deliberately made unreadable. That changes the job: you are almost never recovering missing information, you are undoing a transformation. Identify the transformation first, reverse it with the tool built for it, and fall back to hand-written AST passes only for the custom layer no off-the-shelf tool knows.

When to Use

  • A page or extension ships a large minified/bundled .js you need to read
  • Code is full of _0x1234 identifiers, a big string array, and a rotation IIFE at the top — the obfuscator.io signature
  • A .js.map sourcemap is reachable, or webpack left //# sourceMappingURL
  • You must locate where a request signature, HMAC, token, or crypto key is computed inside a running bundle
  • Reproducing a client-side algorithm (an anti-bot signal, a licence check, a paywall gate) from the shipped code

When NOT to Use

  • React Native / Hermes mobile bundles (index.android.bundle, main.jsbundle, Hermes magic bytes) — use reversing-react-native-apps.
  • A custom bytecode VM or virtualized-JS obfuscator where control flow is interpreted, not just flattened — that is devirtualization; use unpacking-protected-binaries for the VM-lifting approach.
  • Malicious JS in a compromised site or npm package where the goal is IOCs and behaviour, not readability — use hunting-web-backdoors for planted web payloads and auditing-supply-chain for a malicious package.
  • Testing the web app itself (the endpoints the JS calls, XSS, auth) — use testing-web-applications; come back here only to recover a client-side algorithm it needs.
  • WASM modules loaded by the page — that is analyzing-binaries territory.

Identify the Layer Before Touching a Tool

Running the wrong deobfuscator produces plausible garbage. Read the first few hundred bytes and classify:

What you seeTransformationReverse it with
Short names, no whitespace, readable stringsMinification onlyAn unminifier / prettier + rename
webpackChunk, a module map {123: function(e,t,n){…}}Webpack/Rollup bundlingUnbundle to per-module files
//# sourceMappingURL=… or a reachable .mapNothing — the source is right thereSourcemap recovery
_0x hex names + one big string array + a rotation IIFEobfuscator.io string-arraywebcrack / synchrony / restringer
Nested ternaries, while(true){switch(_0x..)} dispatcherControl-flow flatteningAST pass to relink the switch
debugger in a setInterval, self-defending functionAnti-debug / self-defenseStrip the guard before other passes

Most real bundles are layered: webpack on the outside, obfuscator.io on a few modules, a hand-rolled string cipher on the one function that matters. Peel outermost first.

Sourcemaps: Try This First, Always

A reachable sourcemap ends the job before it starts — it contains the original, pre-transform source. Check for it every time, because a large fraction of "obfuscated" production bundles ship or leak one:

  • The //# sourceMappingURL= comment at the bundle's end, and the sibling .js.map even when the comment was stripped (try <bundle>.map).
  • Webpack dev artifacts and source in webpack:// paths inside the map.
  • Recover files with a sourcemap consumer — unwebpack-sourcemap or a short source-map script walks sourcesContent back to a directory tree.

Treat a leaked production sourcemap as a finding in its own right when you are assessing the app, not just a convenience.

Unbundling and Deobfuscation

webcrack is the first tool for a bundle, because it does three of the layers at once: unminify, unpack webpack/browserify into per-module files, and undo obfuscator.io string-array and control-flow obfuscation. Run it, then read the module tree it produces rather than the single blob.

For a bundle that is only obfuscator.io, a dedicated deobfuscator is often cleaner: synchrony or restringer both resolve the string array, reverse the rotation, inline the decoder calls, and flatten the trivial control flow. Compare their output — they fail on different edge cases.

When no tool fully handles the custom layer, write an AST pass with Babel: parse to an AST, @babel/traverse to find the pattern (a specific decoder call, a constant-folded expression, the flattening dispatcher), transform the nodes, and regenerate. This is the durable skill — obfuscators change, but string-array decoding, constant folding, and dead-branch elimination are the same AST operations every time. Prototype the matcher in AST Explorer against the real code before scripting it.

Two failure modes to expect:

  • A deobfuscator that runs the code to "evaluate" the string decoder can execute a payload. For untrusted or malicious bundles, do the string resolution statically or in an isolated sandbox — never eval an unknown bundle on your host to read it.
  • Self-defending / debug-protection re-obfuscates or spins a debugger loop when tampered with. Strip the guard function first (delete the node, or override setInterval) or every downstream pass fights it.

Finding a Routine in a Live Bundle

When static reading is slow — a signing function buried in a megabyte of modules — drive the running page with Chrome DevTools instead:

  • Break on the behaviour, not the code. XHR/fetch breakpoints stop when the signed request fires; the call stack then walks straight back through the signing function. DOM and event-listener breakpoints do the same for UI-triggered logic.
  • Pretty-print in the Sources panel ({}) makes a minified function steppable without any offline work.
  • Hook property access from the console — Object.defineProperty or a Proxy on the object whose method computes the value logs every call and argument with a stack trace, so you see inputs and outputs without reading the math.
  • Override the file with a local, prettified, console.log-instrumented copy (DevTools Local Overrides) to watch the real values flow at runtime.

Runtime observation and static reading are complementary: DevTools tells you which function matters; the AST work tells you what it computes so you can reproduce it offline.

Scope and Authorization

Reading and deobfuscating a bundle you lawfully retrieved is analysis. Two edges need care. Reproducing and replaying a request-signing routine against the origin — the usual reason to reverse an anti-bot signal — acts against that service and needs the same authorization as any other testing; keep it to systems you are permitted to test, and see testing-web-applications. And a sourcemap or bundle pulled from a third-party site is that party's code: recovering it for a security assessment you are engaged to do is fine, lifting a proprietary algorithm for reuse is a different matter.

Rationalizations to Reject

  • "It's obfuscated, so it's slow manual work." Classify first — a reachable sourcemap or a clean webcrack run turns hours into minutes. Hand-reading is the last resort, not the first move.
  • "webcrack's output has a weird spot, so it failed." Layered obfuscation means one tool clears the outer layers and leaves the custom inner one. That residue is the interesting function; switch to an AST pass on it, don't restart.
  • "I'll just eval the string-decoder to get the plaintext." On an untrusted bundle that runs attacker code on your host. Resolve strings statically or sandboxed.
  • "The variable names are gone, so the logic is gone." Minification and obfuscation rename and reshape; they do not delete. Unlike a stripped native binary, the full program is present — you are undoing a transform, not reconstructing lost information.
  • "DevTools is for debugging, not reversing." A fetch breakpoint plus a property hook locates a signing routine faster than reading the bundle, and gives you live inputs and outputs for free.

References

  • reversing-react-native-apps — Hermes/RN mobile JS bundles
  • unpacking-protected-binaries — VM-based / virtualized obfuscation
  • hunting-web-backdoors — malicious planted web JS
  • testing-web-applications — testing the endpoints a recovered algorithm calls
  • reviewing-cryptography — once a client-side crypto/signing routine is recovered

Frequently asked questions

What does the Reversing Obfuscated Javascript AI skill do?

Reverse engineer minified, bundled, and obfuscated browser/Node JavaScript — unpacking webpack chunks, recovering source from sourcemaps, undoing obfuscator.io string-array and control-flow obfuscation with webcrack/synchrony/restringer, and locating a signing or crypto routine in a live bundle via Chrome DevTools. Use when a page ships a huge minified bundle, when code is full of _0x hex identifiers and a rotated string array, when a .js.map is reachable, or when you must find where a request signature or token is computed.

Why use Reversing Obfuscated Javascript on TypingMind?

Because you install it once and use it with any model. Reversing Obfuscated Javascript 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 Reversing Obfuscated Javascript in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/trilwu/secskills/tree/main/secskills-core/skills/reversing-obfuscated-javascript. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Reversing Obfuscated Javascript?

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 Reversing Obfuscated Javascript?

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

Is the Reversing Obfuscated Javascript AI skill free?

Yes. It is published on GitHub by trilwu 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 👇