CVE-2026-101907

Think of this like a mail-forwarding rule you set up saying 'never forward my mail anywhere else.' This flaw means that rule gets ignored by one part of the Axios tool — mail (web requests) gets silently forwarded to places you explicitly tried to block, including private, internal-only addresses inside your own company network.

This matters to you if your website, app, or backend service uses the Axios library's fetch adapter (versions 1.17.0 through 1.20.0) and depends on blocking HTTP redirects as a security safeguard. It is most relevant if your service makes outbound web requests on behalf of users or other systems and has internal-only endpoints that should never be reachable from the outside — this is a risk mainly when such internal endpoints exist and the redirect-block setting was being relied on as protection.

This flaw lets attackers quietly defeat a safety control that was specifically designed to stop requests from reaching internal systems, so any service depending on that control is exposed to internal network probing right now. There's no confirmed exploitation or public exploit code yet, but there's also no fixed version available to install, and the issue is getting sustained security-community attention — treat this as high priority rather than wait for a patch to show up.

1. Confirm exposure: search your codebase and dependency tree for Axios versions in the 1.17.0–1.20.0 range, and identify any call sites that configure maxRedirects: 0 while using the fetch adapter (as opposed to the default XHR/http adapter, which is not affected by this specific bypass).
2. No official fixed version is confirmed in current findings — do not assume 1.20.0 resolves this; monitor Axios's GitHub releases/advisories directly for the actual patched release before downgrading or upgrading based on unverified version numbers.
3. Workaround while unpatched: avoid relying on maxRedirects: 0 via the fetch adapter as a security boundary. Switch affected calls to the default Node http/XHR adapter (which correctly enforces maxRedirects), or implement manual redirect interception/validation (e.g., a custom fetch wrapper that inspects Location headers before following them).
4. Harden the target side: ensure internal-only endpoints enforce their own authentication/network-layer restrictions (e.g., not reachable from the request-originating service's network segment) so a redirect bypass alone isn't sufficient to reach them.
5. No CISA KEV listing or due date applies currently. Monitor: Axios security advisories, your dependency scanner/SBOM for the affected range, and any internal logging for unexpected outbound requests to internal IP ranges or metadata endpoints (classic SSRF indicators).

Weakness: CWE-441 (Unintended Proxy/Intermediary) and CWE-601 (Open Redirect), manifesting as a redirect-based SSRF in Axios's fetch adapter. Affected range: 1.17.0 ≤ version < 1.20.0, per the vendor description (the PATCH finding itself reports no confirmed fix availability, so treat the 1.20.0 fix claim as unverified pending official advisory confirmation). Exploitation status: not in CISA KEV, no public exploit code identified. Predicted exploitation likelihood is low per EPSS (flat trend), consistent with no observed in-the-wild activity. Attack vector requires the application to use the fetch adapter with maxRedirects: 0 as a trust boundary and to make requests whose target can be influenced or redirected by an attacker-controlled response; impact is internal network reconnaissance/access (SSRF), not direct RCE. Press/pulse attention is sustained with a named threat-actor reference (ShinyHunters) in roundup coverage, but this reflects media attention, not confirmed exploitation.