What it does
The scanner looks through the text you paste for strings that have the shape of a secret: cloud and API keys, tokens, private keys, passwords in URLs and secrets assigned to variables. Each finding shows its type, line and column, a masked preview and what to do about it. You can copy the text with every finding replaced by [REDACTED:type], or copy the findings as JSON with masked values only.
What it checks
- AWS access key IDs (AKIA or ASIA) and AWS secret keys written next to a name such as
aws_secret. - GitHub tokens (
ghp_,gho_,ghu_,ghs_,ghr_,github_pat_) and Slack tokens (xoxb-,xoxp-,xoxa-,xoxs-). - Stripe live keys, Google API keys, Anthropic keys (
sk-ant-) and OpenAI style keys (sk-proj-,sk-). - JSON Web Tokens, PEM private key blocks and passwords inside URLs such as
scheme://user:password@host. - Assignments such as
password,secret,token,api_keyoraccess_keyset to a string of 8 or more characters. - High-entropy strings (20 to 200 characters of base64 or hex alphabet with Shannon entropy of 4.0 or more). These are always marked low confidence.
Obvious placeholders (your_api_key, example, xxxx, 0000, changeme and similar) and any line containing dte-ignore are skipped.
What it cannot catch
- Whether a finding is real or still active: the scanner only checks the shape and never contacts any service.
- Secrets with no recognisable shape or prefix, short passwords, secrets split across lines or encoded in another way (base64 of a key, for example).
- Random-looking strings that are not secrets (some IDs and hashes) can still be reported as low confidence, and real secrets that contain a long run of lowercase letters can be missed by the high-entropy check. Hex strings rarely reach the entropy threshold.
- A clean result is not proof that the text is safe. Do not rely on this tool as your only protection.
If you find a secret
Rotate this key. Deleting it from code does not make it safe if it was ever shared or committed. Revoke it at the service that issued it, create a new one and check for unexpected use.
Masking
Previews show the first 4 and last 2 characters of values of 16 or more characters. Shorter values show only their first 2 characters, and a private key shows only its BEGIN line. Switching on "Show values" displays the full text on screen only.
Limits
Text up to 2,000,000 characters is scanned. The table lists the first 500 findings; redaction and the JSON always cover all of them.
What a scan looks like
The built-in example is a made-up .env file. The scan finds six possible secrets, and values are masked in the results unless you switch on “Show values”:
- AWS access key ID (line 2) and AWS secret access key (line 3), both with high confidence.
- GitHub token (line 4), high confidence.
- Password in a URL inside a database connection string (line 5), high confidence.
- Secret assigned to a variable such as
db_password(line 6), medium confidence. - High-entropy string (line 10), low confidence. It looks random but may be a hash, an ID or encoded data.
Obvious placeholders such as your_api_key_here and lines marked with dte-ignore are skipped. The redacted copy replaces each value with a label such as [REDACTED:aws-access-key-id], which is safer to paste into a chat, a ticket or an AI assistant.
What to do when a secret is found
Rotate it. Deleting a key from code does not make it safe if it was ever shared or committed, because copies may exist in Git history, logs and backups. The advice next to each finding is specific to the kind of secret, for example creating a new AWS access key and deactivating the old one, or revoking a GitHub token in your settings.
Frequently asked questions
Does this replace Gitleaks or TruffleHog?
No. It scans the text you paste. It does not look through Git history or whole repositories, which is what those tools are built for. It is a quick check before you share a log, a config file or a snippet.
Why is something flagged that is not a secret?
Low-confidence findings are random-looking strings, and that includes hashes, IDs and encoded data. They are included by default; untick “Include low-confidence findings” to see only the more certain ones.
Is my text uploaded?
No. It is scanned on your device and is not uploaded, stored, put in the URL or logged.
Can I export the findings?
Yes. Copy the redacted text, or copy the findings as JSON with the values masked.
Privacy
This tool runs entirely in your browser. What you enter is processed on your device and is not sent to a server. The page does not write your input to local storage, cookies or the URL.