Zulip security

Zulip’s security strategy covers all aspects of our product and business. Making sure your information stays protected is our highest priority.

Security basics

All Zulip clients (web, mobile, desktop, terminal, and integrations) require TLS encryption and authentication over HTTPS for all data transmission between clients and the server, both on LAN and the Internet. Encryption-at-rest is available with Zulip on-premise, via hardware and software disk encryption of the database and other data storage media.
Zulip’s on-premise offerings can be hosted entirely behind your firewall, or even on an air-gapped network (disconnected from the Internet).
Every Zulip authenticated API endpoint has built in rate limiting to prevent DoS attacks.
Connections from the Zulip servers to Active Directory/LDAP can be secured with TLS. If Zulip is deployed on multiple servers, all connections between parts of the Zulip infrastructure can be secured with TLS or SSH.
Message content can be excluded from mobile push notifications, to avoid displaying message content on locked mobile screens, and to comply with strict compliance policies such as the USA’s HIPAA standards.
Zulip operates a HackerOne bug bounty program to reward hackers for finding and responsibly reporting security vulnerabilities in Zulip. Our completely open source codebase means that HackerOne’s white-hat hackers can audit Zulip for potential security issues with full access to the source code.

Configurable access control policies

Zulip supports private messages (to one or more individuals), private streams with any number of subscribers, as well as public streams available to all organization members. We also support guest accounts, which only have access to a fixed set of streams, and announcement streams, where only organization owners and administrators can post.
By default, users can maintain their own names and email addresses, but Zulip also supports restricting changes and synchronizing these data from another database (such as LDAP/Active Directory).
Zulip provides many options for managing who can join the organization, supporting everything from open to the public (e.g. for open source projects), to requiring an invitation to join, to having an email from a list of domains, to being a member of a specific organization in LDAP/Active Directory.
Zulip can limit the features that new users have access to until their accounts are older than a configurable waiting period.
Zulip also supports customizing whether non-admins can create streams, invite to streams, add custom emoji, add integrations and bots, edit or delete messages, and more.

Authentication

Zulip supports integrated single sign-on with Google, GitHub, SAML (including Okta), AzureAD, and Active Directory/LDAP. With Zulip on-premise, we can support any of the 100+ authentication tools supported by python-social-auth as well as any SSO service that has a plugin for Apache.
Zulip uses the zxcvbn password strength checker by default, and supports customizing users’ password strength requirements. See our documentation on password strength for more detail.
Users can rotate their accounts’ credentials, blocking further access from any compromised Zulip credentials. With Zulip on-premise, server administrators can additionally revoke and reset any user’s credentials.
Owners can deactivate any user, bot, or integration. Administrators can also deactivate any user, bot, or integration except owners.
With Zulip on-premise, session length and idle timeouts can be configured to match your organization’s security policies.

Integrity and auditing

Zulip owners and administrators can configure users’ ability to edit or delete messages, and whether deleted messages are retained in the database or deleted permanently. Zulip by default stores the complete history of all message content on the platform, including edits and deletions, and all uploaded files.
Zulip’s server logging has configurable log rotation policies and can be used for an end-to-end history of system usage.
Zulip stores in its database a permanent long-term audit log containing the history of important actions (e.g. changes to passwords, email addresses, and stream subscriptions).
Zulip’s powerful data exports (on-premise, cloud) can be imported into third-party tools for legal discovery and other compliance purposes. Zulip’s enterprise offerings include support for integrating these with your compliance tools.
Zulip supports GDPR and HIPAA compliance.

The little things

Many products talk about having great security and privacy practices, but fall short in actually protecting their users due to buggy code or poor operational practices.

Our focus on security goes beyond a feature checklist: it’s a point of pride. Zulip founder Tim Abbott was previously the CTO of Ksplice, which provided rebootless Linux kernel security updates for over 100,000 production servers (now the flagship feature of Oracle Linux).

Here are some security practices we’re proud of, all of which are unusual in the industry:

The Zulip server’s automated test suite has over 98% test coverage, including 100% of Zulip’s API layer (responsible for parsing user input). It is difficult to find any full-stack web application with as complete a set of automated tests as Zulip.
Zulip’s Python codebase is written entirely in statically typed Python 3, which automatically prevents a wide range of possible bugs.
All access to user data (messages, streams, uploaded files, etc.) in the Zulip backend is through carefully-audited core libraries that validate that the user who is making the request has access to that data.
Only a small handful of people have access to production servers or to sensitive customer data.
Our error handling systems have been designed from the beginning to avoid including user message content in error reports, even in cases where this makes debugging quite difficult (e.g. bugs in the message rendering codebase).
Zulip has a carefully designed API surface area of only about 100 API endpoints. For comparison, products of similar scope typically have hundreds or even thousands of endpoints. Every new API endpoint is personally reviewed for security and necessity by the system architect Tim Abbott.

These security practices matter! Slack, the most popular SaaS team chat provider, has needed to award hundreds of bounties for security bugs found by security researchers outside the company.