Private by design

Paste a puzzle. Explore a reading.

Runs on this device

Paste text and press Solve. Ctrl / ⌘ + Enter works too.

Advanced settings & examples

Auto-detect ranks English-looking candidates, not probabilities. A selected method never silently falls back to a different cipher.

Limits & verification boundaries

The browser tries selected classical methods, English statistical scoring and encoding layers. It also recognizes PK1–PK10 by exact normalized corpus equality. It cannot break properly implemented AES/RSA or prove an intended plaintext by a high score. Many wiki methods are only reference demonstrations here; their key-recovery attacks require the full version.

A supplied-key reading is arithmetic under your assumption, not independent confirmation. A worked-example reference is labelled as a build-time lookup, never a newly recovered answer. Rankings show actual returned candidates from a bounded search, which may contain fewer than five.

No analysis upload or account. Nothing is saved unless you choose Save below. This hosted page includes third-party advertising; use a locally served/offline copy for sensitive work.

Solver interface improvements

  • Mobile-first input and solve action, with compact navigation.
  • Simple by default; settings, candidate rankings, statistics and manual controls expand on demand.
  • Every cipher article has a Try this cipher action with its real method and full worked example.
  • Optional named investigations let you save, compare and restore attempts on your own device.
  • Explicit live-operation/reference labels, snapshot boundaries and no automatic text saving.
Optional local investigations — save, compare & restore

Explicit device-only storage, no account or cloud sync. Saved snapshots contain ciphertext, plaintext and keys; browser storage is not encrypted. Deleting browser data removes them. Restore does not run a new search or independently verify an old result.

No automatic text or result saving.

The shape of a crypto challenge

Introductory CTF crypto is mostly recognition. The underlying operations are simple — base64, hex, XOR, a classical cipher — and the difficulty comes from not knowing which ones were applied, in what order, and how many times. Experienced players recognise the shapes instantly. This tool does the recognising for you and then does the work as well.

Identifying encodings by shape

Base64

Alphanumerics plus + and /, length a multiple of four, often ending in one or two = padding characters. Mixed case with occasional digits is the tell.

Hex

Only 0-9 and a-f, always an even number of characters. If you see a long string that happens to contain no letters past f, it is hex and not a cipher.

Binary and decimal

Runs of 0 and 1 in groups of eight, or space-separated numbers in the 32 to 126 range — the printable ASCII window. Both are common obfuscation for text that is otherwise in the clear.

XOR

XOR is its own inverse, which is what makes it convenient and what makes it breakable. With a single-byte key there are 255 candidates; the solver tries every one, discards those producing non-printable bytes, and scores the rest against English. With a repeating multi-byte key the structure is identical to Vigenère: find the key length by looking at the index of coincidence of each byte position, then solve each position independently.

Working the layers

The solver treats decoding as a search tree rather than a fixed pipeline. At each node it tries every direct cipher, and separately tests whether the text is a valid encoding it can unwrap. Unwrapping produces a child node and the process repeats. The answer returned is the leaf with the highest confidence, reported with the complete chain that produced it — so you learn the structure of the challenge, not just the flag.

Frequently asked questions

What is single-byte XOR and why is it everywhere in CTFs?

Every byte of the plaintext is XORed with the same one-byte key. There are only 255 keys to try, so it is trivially breakable, which makes it the standard warm-up challenge in introductory CTF crypto categories.

How deep can the layers go?

Both versions peel up to six encoding layers by default. The full version also searches 50 ciphers, handles cipher-on-cipher stacks and scores in six languages, which is what you want for harder challenges.

Can it handle flag formats?

Yes, incidentally — flags like ctf{...} are usually surrounded by enough English or structured text for the scoring to lock on. Very short flag-only inputs are harder because there is little statistical signal.

Why does it sometimes pick the wrong layer to unwrap?

Hex and base64 character sets overlap, so a string can be validly interpretable as either. The solver tries both branches and keeps whichever produces the more English-like result rather than committing to the first guess.

This tool is free and has no account, no upload, no tracking of your text

Everything runs in your browser. If it saved you time, keeping it alive costs nothing but a click.