OpenCraft Security Policy¶
As a professional services company, ensuring the security of our clients' information and the privacy of their users' data is critically important. This security policy states the best practices, expectations, and requirements that we expect each team member to adhere to based on their role.
Our security practices have these goals:
- To ensure that confidential, private, secret, and personal data is not disclosed inappropriately
- To ensure the privacy of our clients and our clients' users
- To ensure that infrastructure we maintain is not accessed nor exploited by unauthorized parties, e.g. tampering with account information
- To ensure that infrastructure we maintain is not utilized to attack users or third parties, e.g. cross site scripting used to attack site visitors
- To ensure that infrastructure we maintain is resilient against attacks
2. Roles and Responsibilities¶
Our CTO, Braden MacDonald, is responsible for developing OpenCraft's security policies, for ensuring that we hold ourselves to a high standard, and for coordinating our response to major security issues.
Team Member Responsibilities¶
Each team member is responsible for:
- Avoiding disclosure of confidential information
- Following the security policies defined in this document
- Reporting security vulnerabilities they discover (to OpenCraft or to the relevant upstream maintainer)
Information that we get from clients or users is classified as follows:
Public: Public information, open source code, etc. Anything that can be found in a Google search and has been intentionally published by the owner/author.
Confidential: Any information which is not clearly public. This includes information that is intended to eventually be public, but it not yet ready to be published.
Confidential information can be shared with anyone on the OpenCraft team or with the relevant client but must never be disclosed to other clients nor to the public.
Private: Information that is clearly private and would never reasonably be expected to be published publicly, such as user personal information (user names, email addresses, online activity, etc.), system access credentials, or proprietary source code.
Private information sometimes must not even be shared with certain employees of the relevant client; if in any doubt about who can see private information, ask the CTO or CEO via JIRA or email.
Secret: Highly private information granted on a need to know basis. Must be marked as "Secret" and the list of people authorized to view it should be specified on the document/information itself.
Never share secret information with anyone who is not explicitly authorized.
Any information about our clients, their business, their customers/users, their future plans, etc. that is not clearly public should be treated as confidential.
Any OpenCraft-related data stored on a team member's laptop or other devices must be encrypted (use full disk encryption or an encrypted partition/folder).
All of our team member's devices that contain OpenCraft-related data must have some sort of auto-lock enabled (require password/biometric/etc. after certain amount of time).
The operating system and other software used by team members on their own laptop or other devices must be kept up to date with all available security patches, and must follow best practices for securely configuring their choice of operating system. If using Windows laptops/machines, they must have a firewall and up to date anti-virus software.
Team members must use sufficiently random, unique passwords for services/systems that they use for OpenCraft work, and must store the passwords in a safe location, such as a vetted and trustworthy password manager.
Two Factor Authentication¶
OpenCraft employees must configure and use two-factor authentication for any cloud services used by OpenCraft that support two-factor authentication, such as Google Apps and Amazon Web Services.
All publicly accessible websites and web applications set up by OpenCraft must allow HTTPS connections, the default protocol used should be HTTPS, and HTTP connections should redirect to HTTPS.
Team members are only to get access to our Vault, storing our client credentials upon successful completion of their trial period. Until that time, they should obtain credentials from other team members on a strictly as-needed basis.
Sending Sensitive Information to Clients¶
When passwords, API keys, or other short anonymous pieces of sensitive information need to be send to clients, the data must either be encrypted or sent partially by email and partially by https://onetimesecret.com/ . For example, send the username by email and the password via onetimesecret; never include a username and password pair in a onetimesecret.
When private or non-anonymous data (e.g. exports of student data or courseware) needs to be sent to a client, it must be transmitted securely and not sent by email. For example, it could be shared by Google Docs or encrypted and then emailed.
Changes to Security Policy and Deviations¶
Changes to our security policy must be approved by the CEO and CTO.
Deviations from the security policy (i.e. using a less secure approach temporarily for a particular client/project/team member) must be approved on JIRA by either the CEO or CTO. Deviations will only be approved in cases where no other reasonable route is possible.
All of OpenCraft's infrastructure should be designed so that the loss of any major provider (or provider's data center in the case of AWS) should not cause any permanent data loss.
For example, we currently use OVH as an OpenStack provider but we use only standard OpenStack APIs and store complete backups on S3 using Tarsnap, so that if OVH were to be offline for a long period of time or to go out of business, we could move all our instances and the instance manager to any other comparable OpenStack provider.
Following the departure of a team member, OpenCraft staff must complete the defined offboarding procedure in full within ten days of the team member's last day. Any passwords that that team member had access to must be changed.
At least every year, OpenCraft shall conduct an internal security audit. Each audit should verify each policy listed in section 4 of this document, as well as investigating the security of OpenCraft's infrastructure in other ways. The audit should result in a written report and any deficiencies identified should be resolved as soon as possible.
Annual security policy review¶
Every year, the CTO and 1-2 other team members shall review the security policy and propose revisions, which will then be considered for approval by the CTO and CEO.
Reporting a Security Vulnerability¶
If you are aware of a security vulnerability affecting our clients or infrastructure, please:
- Open a ticket on JIRA with priority CAT-1. Put "SECURITY" in the ticket name. Put in as many details as you have, and don't worry about filling in all fields correctly.
- Assign the ticket to the designated sprint firefighter, if possible
- Immediately notify email@example.com and include the Jira ticket number. This will create an OpsGenie ticket and page the appropriate on-call team members.
Reporting a Security Breach¶
If you are aware of a security breach (someone has accessed client/OpenCraft data/systems that they should not be able to access), immediately notify security at opencraft dot com.
Response to a Security Breach¶
The current DevOps on-call team will be responsible for responding to any security breach. The response must mitigate the breach, inform any affected client(s), preserve as much event data as possible for forensic analysis, open an incident report ticket on JIRA, and rotate any compromised credentials/systems. Following any such incident, the CTO will work with the team members involved to complete a root cause analysis and action plan to prevent similar incidents in the future.
Risk Identification and Mitigation¶
All team members are encouraged to:
- Follow security mailing lists & announcements from edX Inc., Django, AWS, OVH, and other relevant organizations.
- Notify the CTO by email of any potential risk to OpenCraft/client/user data or systems, even if that risk is not a formal security vulnerability.
- Document such risks in relevant pull requests and/or JIRA tickets.
- Get approval from the CTO or CEO before merging code / deploying systems with an increased risk of data/system exposure.
- The intention of this is that contracts/insurance can be adjusted as appropriate.