How the attack works
GitHub Enterprise Cloud organizations and enterprises can restrict access to their private resources to a list of IP addresses. Once enabled, the list applies to the web interface, the API and Git, to every authentication method (personal access tokens, OAuth app tokens, SSH keys including deploy keys, and GitHub App tokens) and to every role, owners included. Organization owners manage the organization's list and can add entries of their own; entries on the enterprise list are inherited by every organization. A separate setting adds the addresses of installed GitHub Apps to the list automatically, and enterprises with managed users can extend the restriction to user-owned repositories.
A stolen token or key is of little use from outside the allowed networks, which is why an attacker with owner access changes the list. They can disable it, add their own address or range, or turn on the setting that inherits addresses from installed GitHub Apps and then install an app that brings their addresses with it. Because the list also binds owners, the change usually means the attacker is already working from an allowed address and wants direct access. Each change is a single audit log event, and the entry events carry the changed entry.
What an attacker needs
- Control of an organization owner or enterprise owner account.
- A connection from an address that is currently allowed, because the list applies to owners as well.
Impact
Credentials that were confined to approved networks become usable from anywhere, or from the attacker's range: personal access tokens, SSH keys, OAuth tokens and web sessions. An added entry is quieter than disabling the list and can stay in place long after the intrusion is cleaned up.
Detection rules
GitHub IP Allow List Disabled
Detects the IP allow list being disabled for an organization or an enterprise, user-level enforcement of the list being disabled, or the identity provider based IP allow list for web interactions being disabled. After the change, tokens, keys and sessions are accepted from addresses outside the approved networks.
title: GitHub IP Allow List Disabled
id: 833047d0-8e1b-4823-a813-c32280ccdf80
status: experimental
description: >-
Detects the IP allow list being disabled for an organization or an
enterprise, user-level enforcement of the list being disabled, or the
identity provider based IP allow list for web interactions being disabled.
After the change, tokens, keys and sessions are accepted from addresses
outside the approved networks.
references:
- https://attack.mitre.org/techniques/T1556/009/
- https://docs.github.com/en/organizations/keeping-your-organization-secure/managing-security-settings-for-your-organization/audit-log-events-for-your-organization
- https://docs.github.com/en/enterprise-cloud@latest/admin/monitoring-activity-in-your-enterprise/reviewing-audit-logs-for-your-enterprise/audit-log-events-for-your-enterprise
author: Forestall
date: 2026-11-11
tags:
- attack.defense-impairment
- attack.persistence
- attack.t1556.009
logsource:
product: github
service: audit
detection:
selection:
action:
- ip_allow_list.disable
- ip_allow_list.disable_user_level_enforcement
- ip_allow_list.disable_idp_ip_allowlist_for_web
condition: selection
falsepositives:
- A planned change by the owners, for example moving from GitHub's allow list to the identity provider's conditional access policy, or a short window while a locked-out team is restored. Confirm the change record.
level: high
False positives
- A planned change by the owners, for example moving from GitHub's allow list to the identity provider's conditional access policy, or a short window while a locked-out team is restored. Confirm the change record.
GitHub IP Allow List Entry Added or Changed
Detects an address being added to, changed in or deleted from an IP allow list, and the setting that adds the addresses of installed GitHub Apps to the list being enabled. An added address or range lets credentials be used from a network the defenders did not approve.
title: GitHub IP Allow List Entry Added or Changed
id: 30598cd6-821e-4ef2-9e2a-32bcc6a60bb8
status: experimental
description: >-
Detects an address being added to, changed in or deleted from an IP allow
list, and the setting that adds the addresses of installed GitHub Apps to
the list being enabled. An added address or range lets credentials be used
from a network the defenders did not approve.
references:
- https://attack.mitre.org/techniques/T1556/009/
- https://docs.github.com/en/organizations/keeping-your-organization-secure/managing-security-settings-for-your-organization/audit-log-events-for-your-organization
- https://docs.github.com/en/enterprise-cloud@latest/organizations/keeping-your-organization-secure/managing-security-settings-for-your-organization/managing-allowed-ip-addresses-for-your-organization
author: Forestall
date: 2026-11-11
tags:
- attack.defense-impairment
- attack.persistence
- attack.t1556.009
logsource:
product: github
service: audit
detection:
selection:
action:
- ip_allow_list_entry.create
- ip_allow_list_entry.update
- ip_allow_list_entry.destroy
- ip_allow_list.enable_for_installed_apps
condition: selection
falsepositives:
- Network teams updating office, VPN or CI ranges, and entries that an installed GitHub App manages (the integration field names the app). Compare the ip_allow_list_entry value with your approved ranges and exclude the automation account by actor if the volume is high.
level: medium
False positives
- Network teams updating office, VPN or CI ranges, and entries that an installed GitHub App manages (the integration field names the app). Compare the ip_allow_list_entry value with your approved ranges and exclude the automation account by actor if the volume is high.
Detection logic
Patterns that need more than one event, a baseline or knowledge of your own environment. Each is a hypothesis to build and tune in your SIEM, not a ready rule.
Address added to the allow list and used soon after
An owner who adds an entry for their own network does it to use stolen tokens, keys or sessions from there. The entry event names the address or range, and with source IP disclosure enabled every later event names the address it came from. Activity from a newly added range within hours, by the account that added it or with a token, ties the change to its use, while a range added for a new office is used by many people over days.
- An entry is added or changed:
ip_allow_list_entry.createorip_allow_list_entry.update. Keep the address or range inip_allow_list_entry, theactorand the time. - In the following 24 hours, collect the events whose
actor_ipfalls inside that address or range, with theiractorandprogrammatic_access_type. - Alert when the accounts acting from it include the account that added the entry, or act with a token, and when the range is not on your list of company networks.
Eventsip_allow_list_entry.createip_allow_list_entry.update
LimitsGitHub shows source IP addresses only where IP disclosure is enabled in the audit log settings, and never for actions by bots and automated systems. GitHub does not document the format of the ip_allow_list_entry value, so read a few of your own events before matching ranges. The list of company networks is yours to keep. The 24 hours are a starting point.
Triage
- Identify the owner who made the change (actor) and, for entry events, the address or range involved (ip_allow_list_entry) and whether the entry is enabled (active).
- Compare the address with your approved ranges; for entries added by a GitHub App (integration), check when that app was installed and by whom.
- Look for app installations, token approvals and owner changes by the same actor around the change.
- If unexplained, restore the list, revoke the tokens and keys used from the added address, and reset the actor's credentials and sessions.
Response
Once the attack is confirmed:
- Restore the allow list first, since it is what confines stolen credentials to approved networks. Enable it again, delete the unknown entry, and switch off the setting that inherits addresses from installed GitHub Apps if you do not need it.
- If the address came from an installed GitHub App, treat that installation as part of the incident and remove the app unless someone owns it and can explain it.
- Secure the owner account that made the change, including its password, second factor, sessions, tokens and SSH keys. The list binds owners too, so the attacker was already working from an allowed address, and the machine or network behind that address needs examining.
- Find the accounts that connected from outside the approved ranges while the list was open. If source IP disclosure is enabled for the audit log, the events show the address. If it is not, replace the tokens and SSH keys of the accounts the attacker is known or suspected to control.
- After the cleanup, compare every entry on the organization and enterprise lists with your record of approved ranges. A quiet extra entry is the change most likely to be overlooked.
Mitigation
- Keep the number of organization and enterprise owners small, with phishing-resistant two-factor authentication.
- Keep a record of the approved ranges and review the allow list, including entries added by organizations, against it on a fixed schedule.
- Enable the setting that inherits addresses from installed GitHub Apps only if you need it, and review app installations as carefully as list entries.
- Do not rely on the allow list alone; set token lifetimes and restrict personal access tokens so a stolen credential also expires.
- Alert on every change to the list and its settings.
Frequently asked questions
Does the IP allow list also restrict organization owners?
Yes. Once enabled, the list applies to every role, owners included, and to the web interface, the API and Git. An owner who changes the list is therefore already connecting from an allowed address. In an intrusion that usually means the attacker is working from an approved network and wants direct access from their own.
Does the audit log show which address a stolen token was used from?
Only if source IP disclosure is enabled. By default GitHub does not display the source IP address for events in an enterprise's audit log, and the setting has to be switched on in the audit log settings. Some events show no address even then, for example actions performed by bots or automated systems.
Why is an added entry harder to notice than a disabled list?
Disabling the list is one change that affects everyone. An added address or range is a small edit among routine updates of office, VPN and CI ranges, and it can stay in place long after the intrusion is cleaned up. Compare the `ip_allow_list_entry` value of every entry event with your approved ranges.
An entry was added by a GitHub App. Is that expected?
It can be. A setting adds the addresses of installed GitHub Apps to the list automatically, and the `integration` field of the event names the app. The same setting is an attack path, because an owner can turn it on and install an app that brings the attacker's addresses with it. Check when the app was installed and by whom.
References
- MITRE ATT&CK T1556.009: Modify Authentication Process: Conditional Access Policies
- Managing allowed IP addresses for your organization
- Restricting network traffic to your enterprise with an IP allow list
- Audit log events for your organization
- Audit log events for your enterprise
- Displaying IP addresses in the audit log for your enterprise
- Searching the audit log for your enterprise (GitHub Enterprise Server 3.22, actor_ip)