Skip to content
NewCybercrime Report 2026 · Eight Chapters · A Post Is Not An Incident → Explore The ChaptersCybercrime Report 2026 → Explore

Odin Research / What is / Offensive and testing tools

What Is Responder? The LLMNR and NBT-NS Poisoning Tool Explained

A classic internal network tool that answers Windows name lookups and collects login handshakes, now also seen on public cloud servers: how it works and how to shut it out.

By
Odin by Brandefense
Updated
21 Nov 2026
Reading time
7 min read
Type
Offensive tool (credential capture)

In short

Responder is an open-source Python tool that answers LLMNR, NBT-NS and mDNS name-resolution requests on a local network, then runs rogue SMB, HTTP, LDAP, mail and other servers to capture the NTLM authentication that Windows hosts send. Penetration testers use it widely, and MITRE lists APT28, Lazarus Group and Ember Bear among its users.

Key takeaways

What it is
A GPL-3.0 Python tool by Laurent Gaffié that poisons LLMNR, NBT-NS and mDNS and runs rogue authentication servers. MITRE tracks it as software S0174.
How it works
When a Windows host cannot resolve a name, it asks the local network. Responder answers, the host connects to it, and Responder records the NTLM challenge-response.
Why it matters
Captured NetNTLMv2 responses can be cracked offline or relayed to other systems. MITRE maps this to T1557.001.
What Odin Research saw
31 Responder endpoints on 15 public cloud hosts in Monograph 23. Seven were confirmed; four answered with an NTLM challenge on HTTPS port 443.
What to do first
Disable LLMNR and NBT-NS where you can, deny outbound SMB on 139 and 445, and limit outbound NTLM by policy.

What is Responder?

Responder is an open-source Python tool that answers Windows name-resolution broadcasts (LLMNR, NBT-NS and mDNS) on a local network, then captures the login handshakes that fooled hosts send. Its rogue SMB, HTTP, LDAP, mail and other servers collect NTLM hashes for cracking or relay. Testers and attackers both use it.

The project's README calls it "a LLMNR, NBT-NS and MDNS poisoner with built-in rogue authentication servers." The maintained repository is by Laurent Gaffié, is licensed under GPL-3.0, and is a fork of the original SpiderLabs project. It had about 6,700 stars and 870 forks on GitHub on 21 November 2026.

MITRE ATT&CK lists Responder as software S0174. It maps the tool to two techniques: Adversary-in-the-Middle: Name Resolution Poisoning and SMB Relay (T1557.001), and Network Sniffing (T1040). For the wider concept, see adversary-in-the-middle (AiTM).

How Responder works

Windows tries DNS first when it looks up a name. If DNS fails, older fallback protocols ask the local network instead. Responder abuses that fallback.

  1. A lookup fails. A user mistypes a share name, or a stale mapped drive points to a server that no longer exists.
  2. The host broadcasts. It sends an LLMNR, NBT-NS or mDNS query to the local network.
  3. Responder answers first. It claims to be the requested name and points the host at itself.
  4. The host authenticates. It connects to Responder's rogue SMB, HTTP or other server, which asks for NTLM authentication. Windows sends a challenge-response.
  5. The hash is stored. Responder logs the NetNTLMv2 response. MITRE notes this material can be cracked offline or relayed to reach other systems.

The README also lists optional rogue DHCP and DHCPv6 servers and IPv6 router advertisements, which widen the set of hosts it can reach. An analyze mode (-A) lets a tester watch NBT-NS, mDNS and LLMNR requests without answering them. That is useful for scoping a test without touching anyone's traffic.

Responder beyond the local network

Responder is built for internal networks, where it can hear broadcasts. It also works as a plain rogue authentication server on the internet. In that role it pairs with forced authentication (MITRE T1187).

MITRE describes how Windows authenticates on its own when it opens an SMB resource, and falls back to WebDAV when SMB fails. Lure files that reference a remote resource, such as an Office template or a modified shortcut file, can make a host contact an attacker's server. If outbound traffic is allowed, the NTLM handshake leaves the network.

MITRE's mitigation for T1187 is direct: block outbound SMB on TCP 139 and 445 and UDP 137, and filter WebDAV. Strong passwords also make captured hashes harder to crack.

How attackers use Responder

Responder is one of the first tools run after an attacker lands inside a Windows network. It needs no exploit, only a foothold on the right subnet and some patience.

  • State groups. MITRE's S0174 entry links Responder to APT28, Lazarus Group and Ember Bear. MITRE describes it as used to poison name services and gather hashes and credentials on local networks.
  • Relay chains. T1557.001 also lists Impacket, Empire and PoshC2 as procedure examples. Captured authentication is often relayed to another service rather than cracked.
  • Offline cracking. Weak passwords can be recovered from captured hashes with cracking tools such as those managed by Hashtopolis. That turns a hash into a reusable password for lateral movement.

Is Responder legal? Yes, on networks you own or are authorized to test. Running it on other networks is illegal in most countries.

What Odin Research observed

Monograph 23 found Responder exposed to the internet, which is not where the tool normally runs.

  • 31 endpoints on 15 hosts. At 2.07 endpoints per host, Responder had the highest port density in that study. 24 sat in the United States and 4 in Germany, mostly at Namecheap, DigitalOcean and Amazon.
  • Seven confirmed listeners. Four used TCP 443 and answered with HTTP 401 and WWW-Authenticate: NTLM. That gets through firewalls that block outbound SMB but allow HTTPS and WebDAV.
  • Non-web ports. The other three bound TLS sockets on remote desktop (3389), SIP over TLS (5061) and port 5120.
  • Rogue mail hubs. Two suspected Namecheap servers shared one certificate and exposed Responder's full rogue mail and FTP suite on six and seven ports. A mail client that tries plain or NTLM login against them hands over credentials.

The report frames Responder as the credential-capture role in a model alongside JavaChains exploit servers and Ligolo-ng tunnels. No address hosted two of these families.

How to detect Responder

  • NTLM challenges from the internet. Monograph 23 publishes Suricata SID 9523004 for an HTTP 401 with WWW-Authenticate: NTLM from an external server.
  • NTLM leaving the network. SID 9523005 flags an outbound NTLMSSP negotiate message on ports 445, 139, 3389, 5061 or 5120.
  • Responder on your own hosts. A Sigma rule in the report flags Responder.py on Linux at medium level. The Sentinel query joins contact with the confirmed addresses with process events that name it.
  • Poisoning inside the LAN. Look for one host answering many different LLMNR or NBT-NS names, and for authentication to an unexpected internal IP right after a failed DNS lookup.

All rules are on the report page and in the Defense Kit.

How to defend against Responder

  1. Turn off the fallback protocols. MITRE's first mitigation for T1557.001 is to disable or remove the feature (M1042). Disable LLMNR and NetBIOS over TCP/IP where you do not need them.
  2. Stop authentication from leaving. Deny outbound SMB on 139 and 445, restrict the WebClient service, and limit outbound NTLM with the RestrictSendingNTLMTraffic policy, as Monograph 23 advises.
  3. Segment and inspect. MITRE also lists network segmentation (M1030), traffic filtering (M1037) and network intrusion prevention (M1031).
  4. Make cracking expensive. Long passwords and fewer NTLM-only services reduce the value of every captured hash. See Active Directory hardening and credential theft.

Responder compared with similar tools

ToolMain jobMITRE IDNotes
ResponderPoison name resolution and capture NTLMS0174Rogue servers for many protocols
ImpacketWindows protocol toolkit; see our profileS0357Listed under T1557.001
EmpireOffensive frameworkS0363Listed under T1557.001; see Starkiller
PoshC2Offensive frameworkS0378Listed under T1557.001

Frequently asked questions about Responder

What is Responder?

Responder is an open-source Python tool that poisons LLMNR, NBT-NS and mDNS name resolution on a local network and runs rogue authentication servers to capture NTLM hashes.

Is Responder malware?

No. It is a widely used penetration testing tool, licensed under GPL-3.0. Attackers, including groups MITRE names such as APT28 and Lazarus Group, also use it.

Is Responder legal to use?

Yes, on networks you own or are authorized to test. Capturing credentials on other networks is illegal in most countries.

Who made Responder?

The maintained repository is by Laurent Gaffié and is a fork of the original SpiderLabs Responder project.

What is LLMNR poisoning?

When DNS fails, Windows can ask the local network for a name using LLMNR or NBT-NS. A poisoner like Responder answers falsely, so the host connects and sends its NTLM authentication.

What does Responder analyze mode do?

With the -A option, Responder listens to NBT-NS, mDNS and LLMNR requests without answering them. Testers use it to see how much poisoning would be possible.

How do I detect Responder?

Watch for NTLM challenges from external servers, outbound NTLMSSP on SMB, RDP or SIP ports, and hosts answering many name queries. Odin Research publishes Suricata, Sigma and KQL rules in Monograph 23.

How do I stop Responder attacks?

Disable LLMNR and NetBIOS over TCP/IP, block outbound SMB on 139 and 445, filter WebDAV and restrict outbound NTLM. Strong passwords make captured hashes harder to crack.

Sources

  1. Laurent Gaffié, Responder repository, GitHub. github.com/lgandx/Responder (checked 21 November 2026)
  2. MITRE ATT&CK, Software S0174: Responder. attack.mitre.org/software/S0174/ (checked 21 November 2026)
  3. MITRE ATT&CK, T1557.001 Name Resolution Poisoning and SMB Relay. attack.mitre.org/techniques/T1557/001/ (checked 21 November 2026)
  4. MITRE ATT&CK, T1187 Forced Authentication. attack.mitre.org/techniques/T1187/ (checked 21 November 2026)

Related tools and terms

All Odin Research explanations · Brandefense glossary

Odin Research

Get the rules, not just the definition

Every Odin Research report ships with verified indicators and detection rules you can load today.

Open the Defense Kit