In short
Hashtopolis is an open-source client-server tool that splits Hashcat password-cracking jobs across many computers. A central server hands out chunks of work, and agents on GPU machines crack them and report results. Auditors and red teams use it for password audits. Attackers can use it to crack stolen hashes offline, out of sight of the victim.
Key takeaways
- What it is
- A GPL-3.0 project on GitHub for distributed password cracking with Hashcat. It was renamed from Hashtopussy in March 2018.
- How it works
- GPU workers run a Python or C# agent, enroll with the server and keep asking for tasks and keyspace chunks, then send back progress and cracked results.
- Why it matters
- Offline cracking leaves no failed logins on the domain controller. Kerberoasted service tickets are a classic input.
- What Odin Research saw
- 32 Hashtopolis endpoints on 26 hosts in Monograph 19, mostly on ports 8080 and 4200. No hostile cracking job was seen, so all stay on watchlists.
- What to do first
- Remove RC4 from Kerberos, give service accounts long random passwords or gMSAs, and allow Hashcat and Hashtopolis only on named audit hosts.
What is Hashtopolis?
Hashtopolis is an open-source client-server tool for distributed password cracking with Hashcat. A central server splits a cracking job into pieces and hands them to agents running on many GPU machines, which crack their share and report back. Auditors use it for password audits, and attackers can use it on stolen hashes.
The project's README calls it a "multi-platform client-server tool for distributing hashcat tasks to multiple computers". It is published on GitHub under the GPL-3.0 license and had about 1,800 stars on 21 November 2026. The README says it was first called Hashtopussy and was renamed Hashtopolis in March 2018, building on earlier Hashtopus code.
Hashtopolis does not crack anything itself. Hashcat does the cracking. Hashtopolis adds the coordination: task lists, keyspace splitting, agent management and a shared view of results. That turns a room of GPU machines, or a rented fleet, into one cracking cluster.
How Hashtopolis works
The server side is a web application. The README refers to its PHP code and ships Docker Compose files for MySQL and PostgreSQL. Workers run an agent: the README credits a Python agent and a C# agent.
Odin Research's Monograph 19 describes how the worker protocol looks on the wire:
- A GPU worker downloads the agent package, often
hashtopolis.zip. - It enrolls once with a voucher through
POST /api/server.php, then logs in. - It keeps polling with the actions
getTaskandgetChunkto receive work. - It sends results back with
sendProgressandsendCrack.
The server cuts a mask or rule attack into fixed keyspace slices, so many workers can share one job. Monograph 19 lists hash modes that matter in Windows domains, among them Kerberos service tickets (Hashcat mode 13100), AS-REP roasting (18200), Net-NTLMv2 (5600), NTLM (1000) and domain cached credentials (2100).
Why Hashtopolis matters to defenders
MITRE ATT&CK describes password cracking (T1110.002) as recovering usable credentials from captured material such as password hashes. It notes that cracking usually happens offline, on systems the adversary controls.
Offline is the key word. A domain controller sees online password guessing and can lock accounts. It never sees a hash being cracked on a rented GPU rig. Once a weak password is recovered, the attacker logs in as a valid account.
The classic input is Kerberoasting (T1558.003). Any domain user can request service tickets for accounts with a service principal name. MITRE notes that tickets encrypted with RC4 use the service account's password hash as the key, which makes them open to offline brute force.
Monograph 19 models this path from an implant requesting RC4 tickets, through exfiltration, to a Hashtopolis server and its GPU workers. Odin is clear that this is a model built from MITRE ATT&CK, not an observed intrusion. Other inputs include hashes from credential dumping and Net-NTLMv2 captures from tools like Responder.
What Odin Research observed
- 32 endpoints, all watchlisted. Monograph 19 found 32 Hashtopolis endpoints on 26 hosts in at least 10 countries. Russia, Germany and the Netherlands had 5 each.
- Predictable ports. Port 8080 carried the worker API on 14 endpoints, and port 4200 carried the Hashtopolis v2 web console on 12. Six hosts exposed two Hashtopolis ports at once.
- No hostile job seen. Odin saw no hostile cracking job, and no traffic linked these servers to the C2 fleets in the same study. From the outside, an auditor's server and an attacker's rig look the same.
- Busy neighbors. One candidate on port 8443 sat in an Alibaba Cloud range that also held CyberStrikeAI, HexStrikeAI, Metasploit, Sliver, VShell, ARTEX and DNSLog servers. Monograph 17 lists the same kind of candidate in that range. Sharing a cloud range does not show one operator.
- Short watch for home lines. One Hashtopolis listener sat on a Taiwanese subscriber network, where addresses rotate. It got a 72-hour watchlist entry.
The lesson: Hashtopolis servers are context, not indicators. The defense that matters is making stolen hashes worthless.
How to detect Hashtopolis
- Worker polling from inside. Outbound
POST /api/server.phprequests with actions such asregister,getTask,getChunkorsendCrackmean an internal machine is acting as a cracking worker. - Processes and command lines. Alert on
hashcator Hashtopolis agents outside named audit hosts, and on command lines that includehashtopolis.zip,--voucheror Kerberos-related Hashcat modes. - Ticket bursts. On domain controllers, enable auditing of Kerberos service ticket operations (Event 4769). MITRE recommends watching for RC4 (etype
0x17) requests and accounts asking for an unusual number of tickets in a short time. - Logons after dumps. MITRE also suggests looking for valid logons from accounts whose credentials were dumped and that had not logged in before.
Monograph 19 publishes a Suricata signature for Hashtopolis worker polling, a Sigma rule for Hashtopolis and Hashcat execution, a YARA rule for Hashtopolis agents and a Sentinel KQL hunt with an RC4 ticket-burst branch. No rule in that report covers the web console. All are in the Defense Kit.
How to defend against offline password cracking
- Remove RC4 from Kerberos. MITRE recommends AES encryption rather than RC4 where possible. Monograph 19 suggests setting service accounts to AES only.
- Use strong service account passwords. MITRE suggests 25 or more characters for service accounts, or Group Managed Service Accounts. Monograph 19 recommends gMSAs or random passwords of 28 or more characters.
- Limit privileges. Keep service accounts out of groups such as Domain Admins, so one cracked password does not open the whole Active Directory.
- Add MFA. MITRE lists multi-factor authentication as a mitigation, including on external services.
- Control cracking tools. Allow Hashcat and Hashtopolis only on named, isolated audit hosts, and watch everywhere else.
Hashtopolis compared with related credential tools
| Tool | Role | Where it runs | Notes |
|---|---|---|---|
| Hashtopolis | Splits and tracks cracking jobs | Server plus GPU worker agents | 32 endpoints in Monograph 19, all dual-use watchlist entries |
| Hashcat | Cracks hashes | Each GPU worker | The engine Hashtopolis drives |
| Responder | Captures Windows credentials on a network | Attacker host on the network path | Net-NTLMv2 captures are a cracking input |
| Mimikatz | Dumps credentials from memory | Target Windows host | Another common source of hashes to crack |
Frequently asked questions about Hashtopolis
What is Hashtopolis?
Hashtopolis is an open-source client-server tool that distributes Hashcat password-cracking tasks across many computers, usually GPU machines, and collects the results in one web console.
Is Hashtopolis malware?
No. It is a GPL-3.0 open-source tool used for authorized password audits. Attackers can run it too, so an unapproved Hashtopolis agent inside your network deserves investigation.
Is Hashtopolis legal?
Yes, for cracking hashes you own or are authorized to audit. Cracking credentials taken from systems you have no permission to test is illegal in most countries.
What is the difference between Hashtopolis and Hashcat?
Hashcat is the cracking engine. Hashtopolis is the coordinator that splits a Hashcat job into chunks, sends them to many agents and gathers the results.
Was Hashtopolis called Hashtopussy?
Yes. The README says the project was first called Hashtopussy and was renamed Hashtopolis in March 2018.
Which ports does Hashtopolis use?
It depends on the deployment. In Odin Research data, the worker API most often sat on port 8080 and the v2 web console on port 4200.
How do I detect Hashtopolis?
Look for internal hosts posting to /api/server.php with worker actions, hashcat or Hashtopolis agents outside audit hosts, and bursts of RC4 Kerberos ticket requests. Odin Research publishes Suricata, Sigma, YARA and KQL rules for this.
How do I stop Kerberoasting hashes from being cracked?
Disable RC4 for Kerberos, use AES only, and give service accounts long random passwords or Group Managed Service Accounts. That makes a stolen ticket much harder to crack.
Sources
- Hashtopolis, server repository and README, GitHub. github.com/hashtopolis/server (checked 21 November 2026)
- MITRE ATT&CK, Brute Force: Password Cracking (T1110.002). attack.mitre.org/techniques/T1110/002/ (checked 21 November 2026)
- MITRE ATT&CK, Steal or Forge Kerberos Tickets: Kerberoasting (T1558.003). attack.mitre.org/techniques/T1558/003/ (checked 21 November 2026)
Related tools and terms
What is Interactsh?
Interactsh is ProjectDiscovery's open-source OAST server and client for catching DNS, HTTP and other callbacks.
OAST serviceWhat is Burp Collaborator?
Burp Collaborator is PortSwigger's callback service for Burp Suite, used to detect out-of-band vulnerabilities.
OAST serviceWhat is DNSLog?
DNSLog is a family of open-source DNS callback collectors used to confirm blind injection and SSRF flaws.
Offensive toolWhat is ezXSS?
ezXSS is an open-source collector for blind cross-site scripting (XSS) payloads. It records pages where a payload fires.
Interception proxyWhat is mitmproxy?
mitmproxy is an open-source interactive HTTPS proxy. It installs its own root certificate to read encrypted traffic.
Security testing suiteWhat is Yakit?
Yakit is a security testing suite with a desktop interface and a remote engine that can intercept traffic, fuzz web parameters and run exploit checks.
Odin Research
Get the rules, not just the definition
Every Odin Research report ships with verified indicators and detection rules you can load today.