security.txt
security.txt is a small text file, defined in RFC 9116 and served at /.well-known/security.txt, that tells security researchers how to report a vulnerability to the owner of a website or product.
Last updated: November 13, 2026
security.txt is a plain text file that an organization publishes on its website to say how to report a security vulnerability to it. It is defined in RFC 9116, an Informational RFC published in April 2022 by Edwin Foudil and Yakov Shafranovich. A researcher who finds a problem fetches https://example.com/.well-known/security.txt and gets a contact address, a link to the disclosure policy and, optionally, an encryption key, instead of guessing at a generic support address.
Where it goes
- Location:
/.well-known/security.txt, served over HTTPS astext/plainwith a UTF-8 charset. - Legacy path: a file at the top-level
/security.txtis allowed for compatibility and may redirect to the well-known path. If both exist, the well-known one wins. - Scope: the file covers only the exact host it is served from, not subdomains or the parent domain.
www.example.comandshop.example.comeach need their own file or a redirect.
Fields
Each line is a Name: value pair; field names are case-insensitive and lines starting with # are comments. Every web URI in a field must use HTTPS.
| Field | Required | What it holds |
|---|---|---|
| Contact | Yes | A mailto:, tel: or https: address. May repeat; the order shows preference |
| Expires | Yes | Exactly once, as an RFC 3339 date-time. Recommended less than a year ahead |
| Encryption | No | A link to the key for encrypting reports, not the key itself |
| Policy | No | A link to the vulnerability disclosure policy |
| Acknowledgments | No | A link to a page that thanks researchers |
| Preferred-Languages | No | At most once, language tags such as en, fi. English is assumed if it is missing |
| Canonical | No | The URIs where this file is meant to be found; use it with a signature |
| Hiring | No | A link to security job openings |
Other fields are registered with IANA in the security.txt fields registry, and readers must ignore fields they do not recognize. Two matter for product security teams: CSAF, defined by the OASIS CSAF 2.0 standard, which points to a CSAF provider's provider-metadata.json so tools can find your machine-readable advisories (see CSAF), and Bug-Bounty, from the CERT guide to coordinated vulnerability disclosure.
A full example
# Security contact for example.com
Contact: mailto:[email protected]
Contact: https://example.com/security/report
Expires: 2027-10-31T23:00:00.000Z
Encryption: https://example.com/.well-known/pgp-key.txt
Policy: https://example.com/security/disclosure-policy
Acknowledgments: https://example.com/security/hall-of-fame
Preferred-Languages: en, fi
Canonical: https://example.com/.well-known/security.txt
Hiring: https://example.com/careers/security
CSAF: https://example.com/.well-known/csaf/provider-metadata.json
Only the two Contact lines and the Expires line are required. A minimal file is three lines long.
Expires
Expires is required so that a forgotten file stops being trusted. RFC 9116 recommends a date less than a year in the future, and a researcher should treat an expired file as stale. In practice, renew it on a schedule: a calendar reminder or a CI job that fails a month before the date both work. An expired security.txt tells researchers that nobody is reading the inbox.
Signing
RFC 9116 recommends signing the file with an OpenPGP cleartext signature, and using Canonical in a signed file so the signature also authenticates where the file is meant to be found. The signed file wraps the same fields:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256
Contact: mailto:[email protected]
Expires: 2027-10-31T23:00:00.000Z
Canonical: https://example.com/.well-known/security.txt
-----BEGIN PGP SIGNATURE-----
...
-----END PGP SIGNATURE-----
The RFC leaves it to researchers to check that the signing key really belongs to you, so make that easy, for example by publishing the key's fingerprint somewhere other than the same web server.
What security.txt does not do
- It grants no permission to test. The RFC says so explicitly; what researchers may do belongs in your disclosure policy.
- It is not a policy. It points to one. What a policy contains, with a template, is in coordinated vulnerability disclosure.
- It attracts noise. A public address receives spam and automated scanner reports, so route it to a team that triages, not to one person's inbox.
security.txt is for websites and domains. Open source projects on GitHub usually use a SECURITY.md file in the repository for the same purpose.
How this relates to CRACI
CRACI does not receive vulnerability reports or publish security.txt files. Who handles reports and how is covered in PSIRT and coordinated vulnerability disclosure.