Credential detector
Hugging Face access token: format and detection
Short answer
A Hugging Face access token is hf_ followed by 34 letters. Organisation tokens use api_org_ and the same 34-letter body and have their own detector.
What it is
The token grants access to models, datasets and Spaces on the Hugging Face Hub, and can allow writes depending on its scope. The Credentials pack in Maskflare masks it by default.
In Maskflare this is the detector huggingface_access_token, labelled "Huggingface access token", category password or access key, in the Credentials pack.
Format
- The literal prefix hf_.
- 34 letters, either case. Digits are not part of the body in this pattern.
- 37 characters in total.
How Maskflare recognises it
The entropy floor is 2.0 bits per character, so a body of one repeated letter or a short repeating run falls below it.
A keyword prefilter skips any text without hf_.
Why this page shows no sample key
A realistic sample key would look like a live credential to secret scanners and to anyone skimming the page, and an obviously fake one would not pass this detector's length and entropy rules. The format above is enough to recognise the key: the prefix, the length and the character set.
Regex and validation caveats
- The pattern wants the key to be followed by a quote, a backtick, whitespace, a semicolon, an escaped newline or the end of the text. A key directly followed by a comma or a closing bracket is therefore not matched by this pattern.
- hf_ is also a common abbreviation in code (hf_model, hf_token as variable names). Those do not match because the 34 letters must follow the underscore directly.
- Values that look like template variables (${KEY}, {{ key }}, %KEY%), a single repeated letter, true, false or null, and a few file paths are on the detector's allow list and are not masked. A value containing the run abcdefghijklmnopqrstuvwxyz is also treated as an example.
How to detect it in LLM traffic
Tokens appear in notebook cells that log in to the Hub, in Dockerfiles and in training scripts pasted into assistants for debugging.
Maskflare reads the whole request body, including the system prompt, earlier messages and tool arguments, and replaces each match with a token such as mf_mask_d7e0 before the request goes out. Coding agents are covered through hooks that mask tool inputs and results, so a key printed by a shell command does not travel onward in the next prompt. On ChatGPT, browser protection blocks prompts that contain API keys, tokens or private keys outright.
If you build the detection yourself
- Scan the raw text rather than parsed fields: keys sit in code, JSON strings, headers, shell output and pasted logs, often with no label at all.
- Match the prefix and the exact length for the Hugging Face access token. Where the key has no fixed prefix, require the provider's name and an assignment next to the value instead of matching any long string.
- Compute the Shannon entropy of the candidate and reject low values, and keep an allow list of placeholders (${KEY}, repeated characters, the alphabet run), because documentation and templates are full of them.
- Mask the value before the text is sent anywhere. If the key was already sent, treat it as exposed and rotate it; masking later does not recall it.
How Maskflare handles it
- Protect now, Finance - Protect now, Healthcare, Developer and Monitor first: mask (a reversible token).
The available actions are mask (reversible token), redact, partial, hash, format-preserving encryption, generalize, alert and block. Masked values are kept encrypted for 30 days so that answers can be restored, then they expire; prompts and responses are not logged. The full list of detectors is in the detector reference.
Compliance profiles that include it
All 6 compliance profiles include the Credentials pack, and applying them turns this detector on: HIPAA Safe Harbor, GDPR - EU/EEA, UK GDPR, PCI DSS, US personal data and Switzerland - nDSG health data. A profile is a floor: it never turns anything off or weakens a rule that is already stricter.
Frequently asked questions
Does the pattern accept digits in the token?
The catalogue's pattern allows 34 letters after hf_ and nothing else, so a token body with digits would not match this detector.
Are organisation tokens covered?
Yes, by the huggingface_organization_api_token detector with the api_org_ prefix.
Is the Hugging Face access token detector on by default?
Yes. It belongs to the Credentials pack, which the default Protect now template turns on with the action mask.
Does Maskflare store the keys it masks?
Prompts and responses are not logged. A masked value is kept encrypted for 30 days so the answer can be restored, then it expires.
Rules on this page describe what Maskflare's detectors and validators check. They are read from the detection catalogue and are not the specification of the body that issues the identifier. No sample key is shown on purpose.