Tool Decode & Extract
Decoder
Paste any opaque value. It names what it looks like and why, then decodes it: Base64 and other layers, number bases, hashes, JWTs, certificates, SIDs, User-Agents and MAC addresses.
Paste any opaque value and the page tells you what it looks like, and why, before it decodes anything. That covers encoded blobs, numbers, hashes, tokens, certificates, SIDs, User-Agents and MAC addresses. If a value fits more than one shape, it offers the others too.
This runs in your browser. Nothing you type or open is uploaded.
How to read it
Read the reason line. A guess with its reason is easy to check, and a wrong guess is one button away from the right tab.
Peel layers is for obfuscated payloads. It chains Base64, hex, XOR, gzip and so forth, and shows every intermediate step so you can see where it turned readable.
Treat a pasted JWT or certificate like a secret. The page never saves what you paste here, and it asks before putting one in a share link.
Tab by tab
Peel layers
Layered encoding is everywhere in DFIR: a dropped script, a PowerShell one-liner from an EDR alert, a config blob in the registry. It is cheap, reversible obfuscation meant to dodge signatures, and you peel it one layer at a time.
Paste the blob and build a recipe. Each step reads the bytes the step before it produced, so a chain that breaks still shows where and why.
For XOR with no key, leave the key blank. All 256 single-byte keys are tried and the five that give the most printable text are listed. Type the best one into the step to keep going.
Gzip and Deflate use the browser's DecompressionStream. Everything else is a small implementation in this page.
| Step | What it does |
|---|---|
| Base64 decode | Turns Base64 text (A-Z, a-z, 0-9, + and /, often ending in =) back into bytes. The most common outer layer in scripts and droppers. |
| Base64 encode | The reverse, for rebuilding a payload to compare with a sample. |
| Hex decode | Turns pairs of hex digits into bytes, as in shellcode strings and some macro payloads. |
| Hex encode | Shows bytes as hex, useful before searching a hex dump for them. |
| URL decode | Replaces %XX escapes with the characters they stand for, as in phishing links and web logs. |
| URL encode | Escapes characters the way a browser would in a URL. |
| ROT13 | Shifts letters 13 places, so applying it twice gives the original. Windows UserAssist value names use it. |
| ROT47 | Shifts all printable ASCII 47 places, so digits and punctuation change too. |
| XOR (single byte) | XORs every byte with one key. With no key, the tool tries all 256 and ranks the most readable results. |
| Gzip inflate | Decompresses gzip data (starts 1F 8B). PowerShell droppers often gzip a payload before Base64. |
| Deflate inflate (zlib) | Decompresses zlib or raw deflate data, as in PDF streams and some .NET loaders. |
| UTF-16LE decode (e.g. PowerShell -EncodedCommand) | Reads two bytes per character, little-endian. PowerShell -EncodedCommand is Base64 of UTF-16LE, so this is the second step after Base64 decode. |
| Result: Text / Hex | The output of the last step. Hex appears when the bytes look binary, which usually means another layer is left to peel. |
Number bases
Binary, octal, decimal, and hex are four ways to write the same number. Hex wins in computing because one byte is exactly two hex digits, which is much easier to read than eight ones and zeros.
Pick the base your value is in and type. Every other base updates as you go. The math uses big integers, so values past JavaScript's safe-integer limit still convert exactly.
The ASCII row reads the value's bytes two hex digits at a time. If every byte is printable, the number is probably encoded text, which is a quick check on a value pulled from a hex dump or a packet.
| Base | Radix | Digits | Prefix |
|---|---|---|---|
| Binary | 2 | 0 to 1 | 0b |
| Octal | 8 | 0 to 7 | 0o |
| Decimal | 10 | 0 to 9 | none |
| Hexadecimal | 16 | 0 to 9, a to f | 0x |
Hash
A hash proves a piece of evidence has not changed since you collected it. It is also the fastest triage step on an unknown file, because you can check it against a threat feed before you ever open it.
Paste text or choose a file and every digest updates. Text is hashed as UTF-8, so a trailing newline gives a different result than the same text without one.
Paste a known hash into Verify and the tool says which algorithm matched. Paste a bare hash into Identify when you have no context for it. Several algorithms share a length, for example MD5 and NTLM, so Identify lists candidates with a confidence level rather than one answer.
SHA family digests use the browser's Web Crypto API. MD5 is computed by a small implementation checked against RFC 1321's own test vectors.
| Algorithm | Digest | Hex characters |
|---|---|---|
| MD5 | 128-bit | 32 |
| SHA-1 | 160-bit | 40 |
| SHA-256 | 256-bit | 64 |
| SHA-384 | 384-bit | 96 |
| SHA-512 | 512-bit | 128 |
JWT
JWTs show up constantly in investigations: a bearer header in a proxy log, a token pulled from browser storage, an API credential in a config file. This decodes one so you can read who it claims to be and when it expires.
Paste the token. The header names the signing algorithm, and the payload holds the claims. Registered claims such as exp and iat are shown as dates.
It decodes only. Checking the signature needs the issuer's secret or key, and this page has neither. A token that decodes cleanly is not proof that it is genuine.
An alg of none means the token carries no signature at all. A server that accepts one is a finding on its own.
| Claim | Name | Meaning |
|---|---|---|
| iss | Issuer | Identifies the principal that issued the JWT. |
| sub | Subject | Identifies the principal that is the subject of the JWT. The entity the claims are about. |
| aud | Audience | Identifies the recipient(s) the JWT is intended for. A recipient not identified by a value in this claim should reject the JWT. |
| exp | Expiration Time | The expiration time on or after which the JWT MUST NOT be accepted for processing. |
| nbf | Not Before | The time before which the JWT MUST NOT be accepted for processing. |
| iat | Issued At | The time at which the JWT was issued. |
| jti | JWT ID | A unique identifier for the JWT, intended to prevent the token from being replayed. |
Certificate
Turn a PEM or DER certificate into fields you can read and pivot on. Operators rotate C2 addresses freely and keep certificates for months, so the fingerprint and serial are often the better lead.
This decodes. It does not verify. There is no signature check, chain or revocation lookup, so a clean decode can still be self-signed or revoked.
Browsers match the hostname against the Subject Alternative Names, not the CN. Read the SAN list.
Pivot on the SHA-256 fingerprint in Censys, Shodan or crt.sh. A self-signed certificate with an odd subject string has tied together whole families of tooling.
SID and access mask
A Windows access-control entry is a short sentence: a SID naming who, and an access mask naming what they may do. Both read as noise until you split them apart, and both show up in ACLs, in 4663 object-access events, and in the Recycle Bin's folder names.
Paste a SID or a mask, or use a preset. A SID reads S-1-authority-sub-authorities, and the last sub-authority is the RID. RID 500 is the built-in Administrator and 512 is Domain Admins.
A RID of 1000 or above is an ordinary account or group, and only a domain controller can name it. The tool says so rather than guessing.
A mask splits into generic rights in the top bits, standard rights in the middle, and object-specific rights at the bottom. Pick file or registry to read the bottom bits correctly.
User-Agent
A User-Agent is a plain text header the client picks for itself. It is trivially faked, so it is a lead, not an identity. Its real value is spotting a rare, malformed, or inconsistent one in a pile of log lines.
Paste the header value or try a sample. The browser, engine, and OS are best-effort readings of the tokens in the string, and every caveat for this particular string is listed under the table.
Stack-count User-Agents across a proxy or web log and sort ascending. Real browsers collapse into a few common values, so the one-off strings at the bottom are the ones to read.
| Browser | Tokens | Note |
|---|---|---|
| Microsoft Edge (Chromium) | Edg/ (desktop) · EdgA/ (Android) · EdgiOS/ (iOS) | Checked first: Edge UA strings also contain Chrome/ and Safari/ tokens. |
| Firefox | Firefox/ (desktop, Android) · FxiOS/ (iOS) | Excludes Seamonkey/. |
| Chrome | Chrome/ (desktop, Android) · CriOS/ (iOS) | Excludes strings also carrying OPR/ (Opera) or SamsungBrowser/. |
| Safari | Safari/ + Version/ (desktop, or the default "Mobile Safari" on iOS) | Excludes any Chrome/Chromium/CriOS/Edg token. |
MAC address
Every network interface has a 48-bit MAC address, and its structure says a few things before you look up who made the hardware. The first three bytes are the vendor's OUI, and two bits in the first byte say how the address was assigned and who it targets.
Type an address in any common format. The U/L bit says whether it was burned in at the factory or set in software, such as a VM NIC, a VPN adapter, or a phone randomizing its Wi-Fi address. The I/G bit says whether it targets one device or a group.
It does not name the vendor. Look the OUI up in the IEEE registry for that. A locally administered address will not be in the registry at all.
Terms behind it
- Encoding vs. Encryption: Reversible data representation like Base64 that provides no security.
- Obfuscation: Deliberately making code hard to read or analyze.
- JSON Web Token: A signed, self-contained token carrying claims between parties.
- Cryptographic Hash Function: A one-way function mapping any input to a fixed-size fingerprint.
- Public Key Infrastructure: The system of authorities and certificates that binds keys to identities.
Sources
- RFC 4648: The Base16, Base32, and Base64 Data Encodings. The IETF standard for the Base64 encode/decode step this page implements: the alphabet, padding rules, and the URL-safe variant this page's own transform draws from.
- RFC 1952: GZIP file format specification version 4.3. The specification for the gzip container this page's Gzip-inflate step targets, which defines the 1F 8B magic bytes this tool's own step-reference table cites, and the DEFLATE method wrapped inside.
- Unicode Basic Latin code chart. The Unicode Consortium's own official code chart for the Basic Latin block (U+0000-U+007F), listing every control character and printable glyph by exact code point: the authoritative reference behind this page's own printable-ASCII range check (0x20-0x7E).
- FIPS 180-4, Secure Hash Standard (SHS). NIST's authoritative standard for the SHA-1/SHA-2 hash algorithm family this page computes and identifies.
- RFC 7519: JSON Web Token (JWT). The IETF standard defining the JWT claims format this tool decodes: the seven Registered Claim Names (iss, sub, aud, exp, nbf, iat, jti) and the NumericDate format the Expired/Not-yet-valid verdicts are computed from.
- RFC 7515: JSON Web Signature (JWS). The IETF standard for the compact serialization format itself (the three base64url segments this tool splits on), and the header's "alg" parameter this tool reads back without ever using it to verify anything.
- OWASP: Testing JSON Web Tokens. OWASP's Web Security Testing Guide chapter on JWT-specific attack classes (algorithm confusion, the "alg: none" unsecured-JWS trick, weak HMAC secrets), the exact failure modes that make blindly trusting a decoded-but-unverified token dangerous.
- RFC 5280: Internet X.509 PKI Certificate and CRL Profile. The normative definition of every field this page decodes, including the UTCTime two-digit-year window at 50 that silently shifts a certificate by a century when it is read wrong.
- RFC 6125: Service Identity in TLS. Why the Subject Alternative Name rather than the Common Name is what actually authorizes a hostname. Worth reading before trusting a CN.
- [MS-DTYP] §2.4.2: SID structure. Microsoft's own protocol specification for the SID structure: the normative definition of the revision, identifier authority and sub-authority fields this page decodes.
- Well-known SID structures. The authoritative table of fixed SIDs and domain RIDs. Worth reading past the familiar ones: the BUILTIN groups near the end grant more than their names suggest.
- [MS-DTYP] §2.4.3: ACCESS_MASK. How the 32 bits split into generic, standard and object-specific blocks, and why the low 16 bits mean nothing without knowing the object type.
- MDN: HTTP User-Agent header. MDN's reference for the header itself: the product/version/comment syntax, the current per-browser string formats for Firefox, Chrome, Edge, Opera and Safari, and the User-Agent reduction that freezes platform and device values. The primary source behind this page's token tables.
- MDN: Browser detection using the user agent. The guidance page that explains why UA sniffing is unreliable, gives the must-contain/must-not-contain token table this parser's check order follows, and names the "Mobi" substring as the least-unreliable mobile signal. Read this before building anything on top of a parsed result.
- MDN: Firefox user agent string reference. Mozilla's own breakdown of the Gecko UA format. What the legacy "Mozilla/5.0" compatibility token means, why desktop Firefox hardcodes its gecko-trail to 20100101, and the FxiOS quirk where Firefox on iPad carries no Firefox token at all.
- RFC 9110 §10.1.5 HTTP Semantics: User-Agent. The IETF definition of the field this tool parses, including its ABNF and the standard's own warning that user agents generate it with enough detail to be used for fingerprinting, useful when you need to argue from the spec rather than from vendor docs.
- Microsoft Learn: Detecting Microsoft Edge from your website. Microsoft's authoritative source for the Edg / EdgA / EdgiOS token split, why the shortened "Edg" token deliberately differs from legacy EdgeHTML's "Edge", and the confirmation that "Windows NT 10.0" was never updated for Windows 11.
- Chromium: User Agent in Chrome for iOS. The Chromium project's own documentation of the CriOS token and why Chrome on iOS sends an otherwise Mobile-Safari-shaped string: the concrete case for why an iOS browser's engine is WebKit no matter which browser brand the UA claims.
- Chrome for Developers: User-Agent Client Hints. Google's guide to the Sec-CH-UA-* headers and navigator.userAgentData that are replacing UA-string parsing, plus the reasoning behind User-Agent reduction. Worth reading for what a modern client actually exposes beyond the legacy string this tool reads.
- IEEE Registration Authority: Search for an Assignment. The IEEE's own public search tool for OUI/MA-L assignments. Paste the OUI this page decodes into its lookup form to find the organization it was actually issued to, the one question this tool deliberately doesn't answer itself.
- IEEE Public OUI Listing (oui.txt). The Registration Authority's own raw, downloadable listing of every OUI assignment and the organization behind it: the authoritative source behind any vendor-lookup tool, useful for bulk or offline lookups against many OUIs at once rather than one at a time.
Worth using too
- CyberChef. GCHQ's browser-based "Cyber Swiss Army Knife" uses the same drag-and-drop recipe-chaining idea as this tool (Base64, hex, XOR, Gunzip, ROT13…) but with 300+ operations. Reach for it when a payload needs a step this tool doesn't have, like AES/RC4 decryption, regex extraction, or hex-dump parsing.
- dCode.fr: XOR Cipher. Adds automatic XOR key bruteforce (1-16 bytes), key-length cryptanalysis, and frequency-analysis attacks for recovering an unknown key, useful when this tool's single-byte XOR step needs a key you don't already have (dCode's companion ROT-13/ROT-47 pages cover the same rotation ciphers this tool implements).
- de4js. A dedicated JavaScript deobfuscator/unpacker (eval-based packers, JJencode, AAencode, JSFuck, Dean Edwards' Packer) for droppers whose payload is obfuscated JavaScript source itself rather than Base64/hex/XOR/gzip-wrapped binary data, which is outside this tool's scope.
- Binary Hex Converter (binaryhexconverter.com). A free converter that, unlike this page, handles signed values (converting negative decimal/binary numbers via two's complement), the one representation this page's own arithmetic deliberately never attempts.
- VirusTotal. Paste an MD5, SHA-1, or SHA-256 into VirusTotal's free search box to check it against 70+ antivirus engines and community reputation data: the natural next step once this tool computes or verifies a hash and you need to know if it matches known malware.
- Hashes.com Hash Type Identifier. A free, no-signup complement to this tool's own algorithm-guessing feature, covering a much wider range of hash formats (NTLM, bcrypt, and dozens more) for bare hashes that fall outside the MD5/SHA family this tool checks.
- jwt.io Debugger. Auth0/Okta's well-known JWT debugger goes a step further than this page: given the signing secret or public key, it can actually verify a token's signature, not just decode its contents: the check this site's client-side tool deliberately never attempts.
- crt.sh: Certificate Transparency search. The pivot a browser-only decoder cannot do: search CT logs by domain, serial or fingerprint to find every other certificate an issuer has logged.
- PsGetSid (Sysinternals). The other half of the job: resolves a SID to an account name against a live domain, which is exactly what a browser-only decoder cannot do.
- UAParser.js. The widely-used open-source UA detection library, with a far larger maintained pattern set than this page's deliberately small subset: the right tool when you need to parse UA strings in bulk and accept a library dependency (check its licensing, which changed between major versions).
- WhatIsMyBrowser: User Agent Parser & database. A free web-based parser backed by a large searchable database of real-world User-Agent strings: the practical second opinion when a string from a log comes back unrecognized here and you want to know whether anyone else has seen it before.
- Wireshark OUI Lookup Tool. A free, single-box lookup built and maintained by the Wireshark Foundation, wrapping the same IEEE-sourced OUI data in a much friendlier search than the Registration Authority's own multi-field web form: the faster first stop for a one-off vendor lookup once this page has decoded the OUI.