TS-2026-017
Vulnerability from tailscale - Published: Mon, 26 Oct 2026 00:00:00 GMT
Description: HTTP/2 shared serve context could send the wrong Tailscale-App-Capabilities values across mounts.
What happened?
Tailscale Serve can proxy requests to local backends allowing private services to be shared on your tailnet. When enabled with TLS, the server advertises HTTP/2 and all streams on one connection shared a single serveHTTPContext. A race condition occurred because the handler wrote AppCapabilities per request while the reverse proxy read them in a shared hook. Concurrent streams on the same connection could race on this field, causing capabilities from one mount to bleed into requests for another mount.
Tailscale now derives request capabilities from a per-request context rather than sharing state across all streams on an HTTP/2 connection.
This vulnerability is fixed in Tailscale version 1.104.4 or newer.
What was the impact?
Applications that inspect a request's Tailscale-App-Capabilities to authorize a peer might have made incorrect decisions based on capability headers intended for another application. A peer holding a capability for one mount could potentially receive privileged responses from a different mount they are not authorized to access.
Who was affected?
This affects nodes with serve TLS enabled and at least two mounts with different capability requirements on the same node. The issue requires concurrent streams multiplexed on a single connection.
What do I need to do?
If you use Tailscale Serve with TLS enabled, upgrade to Tailscale version 1.104.4 or newer.
Show details on source website{
"guidislink": false,
"id": "https://tailscale.com/security-bulletins/#ts-2026-017",
"link": "https://tailscale.com/security-bulletins/#ts-2026-017",
"links": [
{
"href": "https://tailscale.com/security-bulletins/#ts-2026-017",
"rel": "alternate",
"type": "text/html"
}
],
"published": "Mon, 26 Oct 2026 00:00:00 GMT",
"summary": "\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDescription\u003c/em\u003e\u003c/strong\u003e: HTTP/2 shared serve context could send the wrong \u003ccode\u003eTailscale-App-Capabilities\u003c/code\u003e values across mounts.\u003c/p\u003e\n\u003ch4\u003eWhat happened?\u003c/h4\u003e\n\u003cp\u003e\u003ca href=\"https://tailscale.com/docs/features/tailscale-serve\"\u003eTailscale Serve\u003c/a\u003e can proxy requests to local backends allowing private services to be shared on your tailnet. When enabled with TLS, the server advertises HTTP/2 and all streams on one connection shared a single \u003ccode\u003eserveHTTPContext\u003c/code\u003e. A race condition occurred because the handler wrote \u003ccode\u003eAppCapabilities\u003c/code\u003e per request while the reverse proxy read them in a shared hook. Concurrent streams on the same connection could race on this field, causing capabilities from one mount to bleed into requests for another mount.\u003c/p\u003e\n\u003cp\u003eTailscale now derives request capabilities from a per-request context rather than sharing state across all streams on an HTTP/2 connection.\u003c/p\u003e\n\u003cp\u003eThis vulnerability is fixed in Tailscale version 1.104.4 or newer.\u003c/p\u003e\n\u003ch4\u003eWhat was the impact?\u003c/h4\u003e\n\u003cp\u003eApplications that inspect a request\u0027s \u003ccode\u003eTailscale-App-Capabilities\u003c/code\u003e to authorize a peer might have made incorrect decisions based on capability headers intended for another application. A peer holding a capability for one mount could potentially receive privileged responses from a different mount they are not authorized to access.\u003c/p\u003e\n\u003ch4\u003eWho was affected?\u003c/h4\u003e\n\u003cp\u003eThis affects nodes with serve TLS enabled and at least two mounts with different capability requirements on the same node. The issue requires concurrent streams multiplexed on a single connection.\u003c/p\u003e\n\u003ch4\u003eWhat do I need to do?\u003c/h4\u003e\n\u003cp\u003eIf you use Tailscale Serve with TLS enabled, upgrade to Tailscale version 1.104.4 or newer.\u003c/p\u003e",
"summary_detail": {
"base": "https://tailscale.com/security-bulletins/index.xml",
"language": null,
"type": "text/html",
"value": "\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDescription\u003c/em\u003e\u003c/strong\u003e: HTTP/2 shared serve context could send the wrong \u003ccode\u003eTailscale-App-Capabilities\u003c/code\u003e values across mounts.\u003c/p\u003e\n\u003ch4\u003eWhat happened?\u003c/h4\u003e\n\u003cp\u003e\u003ca href=\"https://tailscale.com/docs/features/tailscale-serve\"\u003eTailscale Serve\u003c/a\u003e can proxy requests to local backends allowing private services to be shared on your tailnet. When enabled with TLS, the server advertises HTTP/2 and all streams on one connection shared a single \u003ccode\u003eserveHTTPContext\u003c/code\u003e. A race condition occurred because the handler wrote \u003ccode\u003eAppCapabilities\u003c/code\u003e per request while the reverse proxy read them in a shared hook. Concurrent streams on the same connection could race on this field, causing capabilities from one mount to bleed into requests for another mount.\u003c/p\u003e\n\u003cp\u003eTailscale now derives request capabilities from a per-request context rather than sharing state across all streams on an HTTP/2 connection.\u003c/p\u003e\n\u003cp\u003eThis vulnerability is fixed in Tailscale version 1.104.4 or newer.\u003c/p\u003e\n\u003ch4\u003eWhat was the impact?\u003c/h4\u003e\n\u003cp\u003eApplications that inspect a request\u0027s \u003ccode\u003eTailscale-App-Capabilities\u003c/code\u003e to authorize a peer might have made incorrect decisions based on capability headers intended for another application. A peer holding a capability for one mount could potentially receive privileged responses from a different mount they are not authorized to access.\u003c/p\u003e\n\u003ch4\u003eWho was affected?\u003c/h4\u003e\n\u003cp\u003eThis affects nodes with serve TLS enabled and at least two mounts with different capability requirements on the same node. The issue requires concurrent streams multiplexed on a single connection.\u003c/p\u003e\n\u003ch4\u003eWhat do I need to do?\u003c/h4\u003e\n\u003cp\u003eIf you use Tailscale Serve with TLS enabled, upgrade to Tailscale version 1.104.4 or newer.\u003c/p\u003e"
},
"title": "TS-2026-017",
"title_detail": {
"base": "https://tailscale.com/security-bulletins/index.xml",
"language": null,
"type": "text/plain",
"value": "TS-2026-017"
}
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.