Al-Khandaq is security software that runs as a privileged service. We take reports about it seriously and thank everyone who reports a problem responsibly.
How to report
- Email: security@akhundbaba.af
- Languages: English, Pashto or Dari.
Please do not open a public issue, post on social media or contact us through donation platforms about a vulnerability.
Include, as far as you can:
- the Al-Khandaq version (About screen, or
khandaq --version) and the operating system; - what an attacker can do, from what starting position (for example: a standard local user, a process in the user's session, someone on the same network, the update server);
- steps or a proof of concept to reproduce it;
- whether you have told anyone else, and when you would like to publish.
Do not attach a diagnostics bundle unless we ask: it contains host names, user names and paths.
What happens next
| Step | Target |
|---|---|
| We confirm we received your report | within 3 working days |
| First assessment (accepted or not, severity) | within 10 working days |
| Fix released for Critical and High issues | within 30 days of confirmation |
| Fix released for Medium and Low issues | in a normal release, normally within 90 days |
| Public advisory | when the fix is released, or at 90 days, whichever is first, unless we agree otherwise |
We are a very small team; if we cannot meet a target we will tell you why and agree a new date with you. We will credit you in the advisory and the changelog unless you prefer not to be named. We do not run a paid bug bounty.
Fixes reach users through the signed self-update channel. Advisories are published on the website's security page and in CHANGELOG.md.
Supported versions
| Version | Security fixes |
|---|---|
| Latest 1.x release | Yes |
| Earlier 1.x releases | No: update to the latest (self-update does this by default) |
| Release candidates and pre-release builds | No |
In scope
- The Al-Khandaq programs: the service (
khandaq-agent), the app, the tray, the command-line tool, the scan worker and the approval helper, on Windows, macOS and Linux, as published by us. - Their local interfaces: the IPC pipe or socket, the approval flow, file and folder permissions of the install and data folders, the installers and uninstallers.
- The update system: signature checks on manifests, releases and content bundles, the timestamp and
key-rotation logic, offline
.kqreleaseand.kqbundleimport. - Licence and organisation-key checks.
- Parsing of untrusted input: scanned files and archives, rule packs, threat lists, update data.
- Our update server and website, for issues that affect users (for example serving altered files). See the limits below.
Examples of what we especially want to hear about: a standard user or a process in the user's
session gaining SYSTEM or root through Al-Khandaq; running code through a crafted file that the
malware scan opens; installing an update or content bundle that is not signed by our keys; a fix or
rollback that changes something other than what was approved; data sent off the computer that
docs/privacy-connections.md does not list.
Out of scope
- Findings that Al-Khandaq reports about your own computer (that is what it is for), or a missed detection or false positive by the malware scan. Please report those as normal bugs.
- Attacks that need administrator or root rights already, or physical access to an unlocked computer, unless they defeat a protection we document (such as the audit log's tamper evidence).
- Vulnerabilities in third-party software (Windows, Defender, winget, apt, YARA Forge rules), unless Al-Khandaq makes them worse. Report those to their vendors.
- The fleet console (
khandaq-fleet), which is a preview and not supported in 1.0. - Denial of service against our servers, volumetric testing, spam, social engineering of our team, and physical attacks.
- Missing security headers, TLS settings or similar on the website with no demonstrated impact.
Safe harbour
If you make a good-faith effort to follow this policy, we will:
- consider your research authorised, and not take or support legal action against you for it, including under anti-circumvention or computer-misuse laws, as far as we are able;
- work with you to understand and fix the problem quickly; and
- publicly acknowledge your contribution if you wish.
Good faith means that you:
- test only on computers and accounts you own or have permission to test;
- do not access, change or keep other people's data, and stop as soon as you reach any;
- do not degrade our servers or other users' service, and do not use automated scanning against our servers beyond what is needed to show the problem;
- give us reasonable time to fix the issue before telling anyone else; and
- do not demand payment in exchange for not publishing.
This safe harbour covers only our software and servers. It cannot authorise testing of systems that belong to others (for example GitHub, the hosting provider or a donation platform). If you are unsure whether something is allowed, ask us first.