AES Full Guide to Encryption: GCM, Password Derivative and Local Example

AES, fix what, don't solve what.

AES (Advanced Enscription Standard) is a group symmetric encryption algorithm: encryption and decryption uses the same key material. Common key length is 128 bit and 256 places(Calls down to AES-128/AES-256 The algorithm itself only defines "how to change the blocks" The real decision is to be made on the safety of the project.

Not Base64. Base64 can decode anyone; AES should not be able to obtain express information without a key (or password). For more details. Base64 Guide with tools Base64 Encoded。

Not Hash. Hash is an irreversible fingerprint and suitable for completeness verification; AES is reversible. Check files for downloads Hash Generators, do not use AES as a "replacement for verification"

Not access control. Encrypted strings can still be copied, forwarded and stored permanently. Whoever gets a secret + weak password can try it violently. HTTPS, authority, audit should still be designed separately.

Not "Put the password in AES, it's safe." User login passwords should be verified using slow-Hashi (bcrypt/scrypt/Argon2, etc.) instead of using AES as a reversible encryption repository to decrypt the code in business at any time, unless you have a clear "restorable key hosting" product demand and compliance program.

GCM Why is this page default

GCM (Galois/Counter Mode) is also availableConfidentialityandIntegrity: If the secret, IV or authentication label is changed when declassified, Web Crypto will fail, not pass out the code. This is a better way to protect the display and draft of "text decryption" in the browser than the old custom of "encrypted without authentication only".

CBC, CTR will still see when compatible with the old system: CBC relies on correct filling and IV management; CTR relying counter/nonce will never repeat. Multi-Model Pages in this Station Text encryption and decryption； Focus on the main tool AES-GCM Single PipelineThe video shows the number of people who have been killed by the attack, and the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, the number of people who have been killed, and the number of people who have been killed.

What do you look like with passwords, salt, IV and secret text?

The script path of this page (consistent with the page security statement) can be summarized as follows:

1. User Input Express and Password, select the length of the key (128/256) and the output format (Base64/Hex).
2. Generate Random salt (16 bytes) and IV (12 bytes)。
3. With Web Crypto:importKey Use the password as a PBKDF2 material deriveKey（SHA-256, about 100,000 inverts) Get AES-GCM keys.
4. crypto.subtle.encrypt({ name: "AES-GCM", iv }, key, utf8Bytes)。
5. Encoding the binary after aggregating in a fixed order:

salt(16B) + IV(12B) + ciphertext_and_tag

Of which GCM authentication labels are usually attached to the end of the secret (dependency) subtle.encrypt , the default tag length, with a 16-bytes scale.The same password, same key length, same output format must be used to decode, and bytes split: the first 16 is salt, the next 12 is IV, the rest is to be decrypted.

Why is it that every time you encrypt the same word, the secret is different?

Because salt and IV Every random time。 This is the expected behaviour:

• The same explicit + same password → The secret string should remain different (to prevent leaks of the same or the same).
• Decrypts are not based on "remember what last secret" but on passwords and packing formats.
• If you see a complete agreement between the two results, you wonder whether you misencoded or copied the wrong field.

The password strength still determines the ceiling.

The cost of a key guess from a password is raised by the PBKDF2.We can't turn "123456" into a military key.。 Recommendations:

• Password length and random reference password generator；
• Passwords and SecretsSubchannel(a) Transmission (e.g., secret e-mail, password face-to-face/telephone/encrypted encryption IM);
• Don't put it on me.PasswordWriteIn.Same one.「CryptoResult.txt」Lee.One.RiseLet's go.。

Encryption declassible step

tools page:AES Encrypted Decryption。 Calculates local completion in browsers; Please. HTTPS or localhost Wait for the security context to use, or else crypto.subtle It may not be available.

Encryption of a paragraph of text (recommended as default)

1. Open tool to confirm top mode as Crypto（Encryption label is active.
2. Yes. Express Text box enter the contents to protect (prefer with non-sensitive examples, e.g., Item目Remarks：Friday.Front只给内网Look.）。
3. Yes. Password Box entry password; You can check whether you have a wrong number and confirm it and then reset it.
4. Key Length Default Selection AES-256； Do not fall to 128 "Talk the Case" without special compatibility needs - this page 128/256 goes GCM, the difference is in key space.
5. Output Format Choose Base64(human pasting, placement in JSON fields more frequently) or Hex(Indicated only 0-9a-f, some log systems are more friendly).
6. Click Crypto。 Below Secret (a) The emergence of a long chain;Error The area should be empty.
7. Points Copy Take the secret. - It's over. Clearand consider cleaning the clipboard.

B. Decrypt texts (trip-to-validation)

1. Switch to Decipher Label: Enter the label to become "secret" and the button to "declassified".
2. Paste the complete secret (do not insert spaces, line breaks, quotation marks without permission, unless you are sure that the transfer layer is added and the other side will remove it - this page decodes by the whole string).
3. FillSame password for encryption；Key length, output format to be encrypted(Base64 secret not to solve with Hex).
4. Points Decipher。 Successfully, the output area is specified in UTF-8; Failure is shown in the error table below.
5. Suggested custom: Once encrypted, do a decryption self-measurement in the same browser and send it to the collaborators.

C. Minimum replicable examples (mental model, non-fixed secret)

Because salt/IV is random,CannotThe blog posts a list of "this word will always be for a string of Base64" and writes in the document. The correct demonstration is the process, not the dead result:

If step 4 fails and step 3 is successful, priority is given to whether the string is cut while still in encryption mode, whether the key length has been changed, and whether the string is cut while copying.

Capacity boundaries (relative to real pages)

• Yes: AES-GCM； AES-128/256； Password + PBKDF2; Random salt and IV; Base64/ Hex output; Encryption/declassification mode switching; Copy, empty, password-show switching; Local Web Crypto.
• None: A key switch compatible with the default parameters of OpenSSL command line.
• Not with Text encryption and decryption The default interconnectivity of the "calendar prefix" - the return and return of the computer on the same page before changing the tool.

Common failures and contrasts

Suggested error order

1. Confirms that the mode is decryption and that the password, key length, output format three packages are encrypted.
2. Confirm that the secret is complete, no additional quotes and AES-GCM: Waiting for outside prefixes (that's the format of another page).
3. The same browser, same page, "encrypted and decrypted" to reduce environmental differences.
4. The failure is not an explicit re-entry of the text, excluding interference with the "original text itself contains invisible characters".

Real work scenarios

Scenario A: Handing the configuration to colleagues without taking an explicit e-mail

The job is to give a temporary link description to a remote colleague. Encryption with this page AES-256 + Base64, secret poster; The password is given by the established voice or enterprise IM alone. The other side used the same page to decrypt.The agreement is to write in:"Thousands of AES encrypted decryption pages, GCM, 256, Base64." Not just "I'm AES once" -- the other side may open multiple mode pages or desktop OpenSSL, the parameters are not right.

Scene B: Local diary/draft against "spatching" and misinformation

(a) Writing incomplete personal notes on shared computers: only secret files are kept after writing encryption; Empty pages before leaving. This reduces the risk of "screens unlocked, file manager previews clear" but does not stop the rivals of the wooden horse and disk — who are expected to be honest.

Scenario C: Training new recruits against "coding vs encryption"

Same sentence demo-secret：

1. Yes. Base64 - Encoding. - Anyone can solve it;
2. In this page, AES encryption → without password should fail;
3. Yes. Hash Do SHA-256 thither.

Three steps in parallel, more effective than speaking of the concept of one hour. Training is conducted using false data only.

scene D: AES encryption field in interface document

Backend says "AES encrypted Base64". First, ask yourself:Mode (GM/CBC), original byte or password, pre-set IV, adhesive or not, intersemination, number of times。 This page is "Palphine + PBKDF2 + Fixed Packing". If the other side is "fixed 16 bytes key + independent IV field", the user will be able to use the word "soft" to use the word "soft" in the context of the "soft" and "show" field.couldn'tI expect this text to be accessed directly. This tool is appropriate for your understanding of GCM behavior on the side of the browser or for protecting your own intercommence notes, rather than pretending to be compatible with all back-end dialects.

Scenario E: Migration from weak programmes

The historical system uses "sensitive fields Base64" or "reversible XOR" as encryption. The migration period can be used to protect the exported migration list from this page, while planning for official encryption at the service level (envelope encryption, KMS, field-level schemes). This page does not cover batch re-entry.

Scene F: Too weak a password to add "no more than no more."

Somebody used it. 111111 Can not encrypt this list of clients before uploading it to the Web. The attackers can be guessed offline after downloading the secret text. Use pre-encrypt password generator Generate high entropy passwords and enable password manager storage; There are passwords available for needs assessment Password Strength(if available).

How to choose relevant tools

Selection tips: (a) The confidentiality of AES (this page) is reversible; To multi-module compatibility test →text encryption page; Textual binary → Base64; Just prints, Hash; The passwords are not random enough for Mr. Scrambling to be a device.

FAQ

Continue reading

A non-sensitive Chinese word goes "encrypted copy decryption decryption hyphenation should fail to clear the password" and then reads the secret structure description, so you can create the correct mind model. Write the team agreement as "Tool URL + GMM + Key Length + Output Format + Password Channel" is much more reliable than the "AES One Time" word.