Thank you for your interest in keeping breaches.cloud up to date. Hopefully, this document covers the details you need to submit a new breach or update an existing one.
This site is for security incidents where a customer's mistake in public cloud configuration was a direct or proximate cause of the incident or breach. Vulnerabilities and issues with the cloud providers themselves belong at The Open Cloud Vulnerability & Security Issue Database.
It's essential to provide as much detail via press releases, indictments, FTC civil actions, or court filings as possible. Technical detail is critical to inform the cloud security community of the measures they must take to avoid being involved in a similar incident.
Avoid speculation where possible; however, it's ok to "fill in the gaps" between details in an indictment and how that detail might have occurred technically in a cloud provider.
Avoid sensational language unless it is part of a news article or press release headline.
If a person or company is named, that person or company must be explicitly named in a sourced document. Traditionally, the United States treats all individuals as innocent until proven guilty. As such, any mention of a suspect must clearly indicate if that person is indicted, convicted, pleaded guilty, or was found not guilty by a court of law.
Any mention of an individual must include a link to the indictment where that person is named.
This site uses Hugo to render the content in markdown to HTML. To create a testing version of this site:
- Fork & clone this repository
- Download the Submodule for the Fuji theme
$ git submodule init $ git submodule update
- Download Hugo Extended v0.111.1. You must download the extended version. Extract the binary into your path and call it
hugox
- Run
make test-preview
- A browser window will open to localhost:1315 and the hugo webserver will start.
You should have a functioning copy of the breaches.cloud site.
To create a new page for a new incident run:
make incident name=INCIDENT_NAME
You will find the new files under content/incidents/INCIDENT_NAME
All pull requests (PRs) will be reviewed to ensure they follow the Editorial Guidelines above. We will ask for changes if the language is sensationalist or provides facts without proper citation. Please make PRs to the preview
branch of https://github.com/primeharbor/breaches.cloud
Please do not disclose facts about an incident you were involved in that might be under NDA. 🚓 Please consult your attorney if you were responsible for an incident and disclosing new facts on a case. :) 🚓
Please complete the FrontMatter of each new incident:
---
title: "Imperva RDS Snapshot"
date: 2023-04-09T16:51:47-04:00
# lastmod: 2023-04-09T16:51:47-04:00
description: "A Cloud Security focused analysis of a data breach at Imperva in October 2018"
# Set to false before publishing
draft: false
# This is not used for SEO, but allows users on the site to view incidents by keyword
tags: ["AWS", "Public Snapshots", "Public Instance"]
showToc: true # Do not change
author: "Chris Farris"
author_url: "https://www.chrisfarris.com/"
# These details are used to populate the Incident Index page.
victim: "Imperva"
incident_start_date: 2018-10-01
incident_end_date: 2018-10-31
disclosure_date: 2019-08-27
final_status: "Closed"
---
title
is what is seen on the front page and in the title tagdate
is the publishing datelastmod
should be updated if you're changing or updating an incidentdescription
is for SEO and is designed to help people find the contenttags
are how we map causes. It's intended to allow a visitor to filter to see all cases where a particular provider or problem led to an incident or breach.author
andauthor_url
is for the original author of a post.victim
,incident_start_date
,incident_end_date
disclosure_date
are used on the Incidents Page and in the details table in the top of each post. Dates should be YYYY-MM-DDfinal_status
should indicate the current status of the case (yes, it is misnamed)
If a company has had more than one incident, please put the year in the title (ie, Chegg (2018)
).
The general format of a page should be a brief introduction, followed by a <!--more-->
tag followed by this outline:
## Incident
### Details of the Incident
### Timeline of the Incident
### Attribution / Perpetrator
### Long-term Impact
## Summary of Coverage
## Cloud Security Lessons Learned
The Long-term impact
section is optional for new incidents, as we do not know what that impact will be. Otherwise, the purpose of this section is to state the costs, penalties, etc, for the victim and people associated with the victim.
The Attribution / Perpetrator
should cite primary sources if an individual or group is named, otherwise state: No known attribution at this time.
The Cloud Security lessons learned
section is where you can highlight how the Perpetrator was able to use cloud mistakes in order to conduct their attack. The goal is to highlight best practices and why they should be followed.
If you are a member of the Cloud Security Forum Slack, you can join the #breaches.cloud channel where questions or discussion can occur.