Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-101913 Overview
The ip-address JavaScript library contains a logic flaw in its IPv6 link-local address classification. Prior to version 10.5.1, the Address6.isLinkLocal method in src/ipv6.ts recognizes only the fe80::/64 prefix instead of the full fe80::/10 range defined by RFC 4291. Attackers can supply an IPv6 address elsewhere within fe80::/10 to bypass trust-boundary checks that depend on isLinkLocal. The same address is inconsistently identified as link-local by the sibling getType and getScope methods, exposing the classification gap. A successful bypass permits access to on-link hosts outside the intended trust boundary. The issue is fixed in ip-address version 10.5.1 [CWE-697].
Critical Impact
Inconsistent link-local classification enables trust-boundary bypass, potentially exposing SSRF filters and internal network protections that rely on isLinkLocal for IPv6 validation.
Affected Products
- ip-address npm library versions prior to 10.5.1
- JavaScript and TypeScript applications using Address6.isLinkLocal for security decisions
- SSRF filters, allow-list checks, and network-boundary validators built on the affected library
Discovery Timeline
- 2026-09-28 - CVE-2026-101913 published to NVD
- 2026-09-30 - Last updated in NVD database
Technical Details for CVE-2026-101913
Vulnerability Analysis
The vulnerability is an incorrect comparison flaw [CWE-697] in the Address6 class. RFC 4291 §2.4 defines IPv6 link-local unicast addresses as the entire fe80::/10 prefix. The affected isLinkLocal implementation narrows this check to fe80::/64, treating any address between fe80:0:0:1:: and febf:ffff:ffff:ffff:ffff:ffff:ffff:ffff as non-link-local.
Applications commonly call isLinkLocal to reject or permit link-local traffic when enforcing trust boundaries. Since getType and getScope correctly classify the full /10 range, callers combining these APIs receive contradictory answers for the same address. Attackers exploit this asymmetry to bypass security logic while still reaching hosts that behave as link-local on the wire.
Root Cause
The root cause is an overly restrictive prefix constant. The pre-patch method matched addresses against fe80::/64 rather than the complete fe80::/10 unicast link-local range. The fix aligns isLinkLocal with RFC 4291 and also recognizes IPv4-mapped or NAT64 addresses whose embedded IPv4 falls within 169.254.0.0/16, such as ::ffff:169.254.169.254.
Attack Vector
An attacker supplies a crafted IPv6 address inside fe80::/10 but outside fe80::/64 to any application that uses Address6.isLinkLocal for filtering. The check returns false, allowing the request to pass validation. The application then processes the address as public or trusted, enabling reach to on-link resources such as cloud metadata endpoints exposed via NAT64 or link-local peers.
// Security patch in src/ipv6.ts (v10.5.1)
/**
- * Returns true if the address is a link local address, false otherwise
+ * Returns true if the address is a link-local unicast address in `fe80::/10`
+ * (RFC 4291 §2.4)
+ * or an IPv4-mapped / NAT64 address whose embedded IPv4 address is link-local
+ * (`169.254.0.0/16`, e.g. `::ffff:169.254.169.254`), false otherwise.
* @returns {boolean}
*/
isLinkLocal(): boolean {
Source: GitHub commit d03e960
Detect Identity Attacks the Moment Authentication Fails
Singularity™ Identity surfaces credential misuse, lateral movement, and privilege escalation in real time, so you can contain the breach before it spreads.
Stop Identity AttacksDetection Methods for CVE-2026-101913
Indicators of Compromise
- Inbound or outbound requests containing IPv6 destinations within fe80::/10 but outside fe80::/64, such as fe80:1::1 or feb0::1
- Application logs where isLinkLocal returns false while getType or getScope classifies the same address as link-local
- Requests to IPv4-mapped NAT64 targets like ::ffff:169.254.169.254 attempting to reach cloud metadata services
Detection Strategies
- Inventory package.json and package-lock.json files across repositories and build artifacts for ip-address versions below 10.5.1
- Add runtime assertions that compare isLinkLocal, getType, and getScope results and alert on disagreement
- Review SSRF and network allow-list code paths that consume Address6 output for boundary decisions
Monitoring Recommendations
- Enable web application firewall rules that reject IPv6 destinations in fe80::/10 when they should not be reachable
- Log all outbound requests to link-local and IPv4-mapped 169.254.0.0/16 addresses from server-side fetchers
- Ship dependency scanner output to a central data lake and alert on any ip-address package below 10.5.1
How to Mitigate CVE-2026-101913
Immediate Actions Required
- Upgrade ip-address to version 10.5.1 or later in all Node.js and browser build pipelines
- Rebuild and redeploy container images that bundle the vulnerable library
- Audit application code that relies on isLinkLocal for security decisions and add server-side network egress controls as defense in depth
Patch Information
The fix is available in ip-address v10.5.1. See the GitHub release notes for v10.5.1 and the GHSA-rpw4-54j3-4h4q advisory. The patch aligns isLinkLocal with the full fe80::/10 range and adds recognition of IPv4-mapped link-local addresses.
Workarounds
- Replace calls to Address6.isLinkLocal with a custom check that validates addresses against the entire fe80::/10 prefix
- Combine getType or getScope results with isLinkLocal and reject the address if any method reports link-local
- Enforce network-layer egress filtering that blocks all IPv6 link-local and IPv4 169.254.0.0/16 destinations for untrusted request contexts
# Upgrade the vulnerable dependency
npm install ip-address@^10.5.1
# Verify installed version
npm ls ip-address
# Audit for known advisories
npm audit --production
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

