GHSA-PQG7-V6WH-3PFP
Vulnerability from github – Published: 2026-07-14 19:46 – Updated: 2026-07-14 19:46Description
The HTTP reverse proxy handler in tsdproxy does not strip the X-Forwarded-For (or X-Real-IP) header from incoming requests before calling r.SetXForwarded(). This allows an authenticated Tailscale user to inject arbitrary X-Forwarded-For values that are forwarded verbatim to backend services.
// internal/proxymanager/port.go -- Rewrite function
Rewrite: func(r *httputil.ProxyRequest) {
r.SetURL(pconfig.GetFirstTarget())
r.Out.Host = r.In.Host
// Strips tsdproxy identity headers (correct)
r.Out.Header.Del(consts.HeaderID)
r.Out.Header.Del(consts.HeaderRemoteUser)
r.Out.Header.Del(consts.HeaderXForwardedUser)
// ... other identity headers deleted ...
// X-Forwarded-For is NOT deleted before SetXForwarded!
// X-Real-IP is NOT deleted at all!
r.SetXForwarded() // APPENDS client IP to attacker-controlled XFF list
},
Per Go's httputil.ProxyRequest.SetXForwarded() documentation:
If the inbound request has an existing X-Forwarded-For header, SetXForwarded appends the inbound request's remote address to the list.
Result when attacker sends X-Forwarded-For: 127.0.0.1: - Backend receives: X-Forwarded-For: 127.0.0.1, - If backend reads first element as "original client", attacker appears as 127.0.0.1
X-Real-IP is not handled at all -- if the attacker sets X-Real-IP: 127.0.0.1, it is forwarded to the backend verbatim without any overriding or stripping.
Many backend applications trust the first element of X-Forwarded-For (or X-Real-IP) for: - IP-based access control (admin panels restricted to 127.0.0.1) - Rate limiting tied to source IP - Audit logging - Geo-blocking or network-segment restrictions
This is particularly impactful in tsdproxy's intended use case where the backend service is only accessible through tsdproxy -- making the proxy's header handling the sole enforcement point.
CVSS
CVSS v3.1: AV:N/AC:L/PR:L/UI:N/S:C/C:L/I:H/A:N = 7.7
Severity
High
Affected Code / Files
internal/proxymanager/port.go-- newPortProxy Rewrite closure- Missing: r.Out.Header.Del("X-Forwarded-For") before r.SetXForwarded()
- Missing: r.Out.Header.Del("X-Real-IP") unconditional strip
Steps to Reproduce
- Deploy tsdproxy with a backend service that restricts /admin to 127.0.0.1 via X-Forwarded-For (e.g., Nginx with real_ip_header X-Forwarded-For and real_ip_recursive on)
- As an authenticated Tailscale user (non-admin), make a request through the proxy:
curl -H "X-Forwarded-For: 127.0.0.1" \
https://<proxy-hostname>.ts.net/admin
- Backend receives: X-Forwarded-For: 127.0.0.1,
- Nginx real_ip_recursive resolves left-most non-trusted IP; if tailscale range is the only trusted range, 127.0.0.1 becomes the "real" IP, bypassing admin restriction.
For the X-Real-IP vector:
curl -H "X-Real-IP: 127.0.0.1" \
https://<proxy-hostname>.ts.net/admin
# Backend receives X-Real-IP: 127.0.0.1 verbatim
PoC Script (if used)
#!/bin/bash
# Demonstrate XFF injection through tsdproxy
PROXY_HOST="${1}" # e.g. myapp.my-tailnet.ts.net
curl -v \
-H "X-Forwarded-For: 127.0.0.1" \
-H "X-Real-IP: 127.0.0.1" \
"https://${PROXY_HOST}/"
# Expected: backend sees XFF: 127.0.0.1, <tailscale-ip>
# backend sees X-Real-IP: 127.0.0.1 (unmodified)
Impact
An authenticated Tailscale user who should only have regular user access can: 1. Bypass IP-based admin restrictions on the backend application by spoofing X-Forwarded-For to 127.0.0.1 2. Appear as any arbitrary IP address in audit logs 3. Bypass rate limiting tied to source IP 4. Bypass geo-blocking or network-segment restrictions enforced by the backend
This is especially impactful because tsdproxy is designed as the sole access point for backend services that are otherwise network-isolated -- making the proxy the only enforcement boundary.
Fix: Add r.Out.Header.Del("X-Forwarded-For") and r.Out.Header.Del("X-Real-IP") in the Rewrite closure before calling r.SetXForwarded(). This ensures only the real Tailscale client IP appears in the XFF chain.
Credits
Reported by Vishal Shukla (@shukla304) using sechub.dev AI Agent
Support
If this disclosure work has been useful, sponsoring helps fund continued open-source security audits -- appreciated either way.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c 3.0.0-alpha.3"
},
"package": {
"ecosystem": "Go",
"name": "github.com/almeidapaulopt/tsdproxy"
},
"ranges": [
{
"events": [
{
"introduced": "0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-74"
],
"github_reviewed": true,
"github_reviewed_at": "2026-07-14T19:46:15Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "## Description\n\nThe HTTP reverse proxy handler in tsdproxy does not strip the X-Forwarded-For (or X-Real-IP) header from incoming requests before calling r.SetXForwarded(). This allows an authenticated Tailscale user to inject arbitrary X-Forwarded-For values that are forwarded verbatim to backend services.\n\n```go\n// internal/proxymanager/port.go -- Rewrite function\nRewrite: func(r *httputil.ProxyRequest) {\n r.SetURL(pconfig.GetFirstTarget())\n r.Out.Host = r.In.Host\n\n // Strips tsdproxy identity headers (correct)\n r.Out.Header.Del(consts.HeaderID)\n r.Out.Header.Del(consts.HeaderRemoteUser)\n r.Out.Header.Del(consts.HeaderXForwardedUser)\n // ... other identity headers deleted ...\n\n // X-Forwarded-For is NOT deleted before SetXForwarded!\n // X-Real-IP is NOT deleted at all!\n r.SetXForwarded() // APPENDS client IP to attacker-controlled XFF list\n},\n```\n\nPer Go\u0027s httputil.ProxyRequest.SetXForwarded() documentation:\n\u003e If the inbound request has an existing X-Forwarded-For header, SetXForwarded appends the inbound request\u0027s remote address to the list.\n\nResult when attacker sends X-Forwarded-For: 127.0.0.1:\n- Backend receives: X-Forwarded-For: 127.0.0.1, \u003creal-tailscale-client-ip\u003e\n- If backend reads first element as \"original client\", attacker appears as 127.0.0.1\n\nX-Real-IP is not handled at all -- if the attacker sets X-Real-IP: 127.0.0.1, it is forwarded to the backend verbatim without any overriding or stripping.\n\nMany backend applications trust the first element of X-Forwarded-For (or X-Real-IP) for:\n- IP-based access control (admin panels restricted to 127.0.0.1)\n- Rate limiting tied to source IP\n- Audit logging\n- Geo-blocking or network-segment restrictions\n\nThis is particularly impactful in tsdproxy\u0027s intended use case where the backend service is only accessible through tsdproxy -- making the proxy\u0027s header handling the sole enforcement point.\n\n## CVSS\nCVSS v3.1: AV:N/AC:L/PR:L/UI:N/S:C/C:L/I:H/A:N = 7.7\n\n## Severity\nHigh\n\n## Affected Code / Files\n- `internal/proxymanager/port.go` -- newPortProxy Rewrite closure\n- Missing: r.Out.Header.Del(\"X-Forwarded-For\") before r.SetXForwarded()\n- Missing: r.Out.Header.Del(\"X-Real-IP\") unconditional strip\n\n## Steps to Reproduce\n1. Deploy tsdproxy with a backend service that restricts /admin to 127.0.0.1 via X-Forwarded-For (e.g., Nginx with real_ip_header X-Forwarded-For and real_ip_recursive on)\n2. As an authenticated Tailscale user (non-admin), make a request through the proxy:\n\n```bash\ncurl -H \"X-Forwarded-For: 127.0.0.1\" \\\n https://\u003cproxy-hostname\u003e.ts.net/admin\n```\n\n3. Backend receives: X-Forwarded-For: 127.0.0.1, \u003cyour-tailscale-ip\u003e\n4. Nginx real_ip_recursive resolves left-most non-trusted IP; if tailscale range is the only trusted range, 127.0.0.1 becomes the \"real\" IP, bypassing admin restriction.\n\nFor the X-Real-IP vector:\n```bash\ncurl -H \"X-Real-IP: 127.0.0.1\" \\\n https://\u003cproxy-hostname\u003e.ts.net/admin\n# Backend receives X-Real-IP: 127.0.0.1 verbatim\n```\n\n## PoC Script (if used)\n```bash\n#!/bin/bash\n# Demonstrate XFF injection through tsdproxy\nPROXY_HOST=\"${1}\" # e.g. myapp.my-tailnet.ts.net\ncurl -v \\\n -H \"X-Forwarded-For: 127.0.0.1\" \\\n -H \"X-Real-IP: 127.0.0.1\" \\\n \"https://${PROXY_HOST}/\"\n# Expected: backend sees XFF: 127.0.0.1, \u003ctailscale-ip\u003e\n# backend sees X-Real-IP: 127.0.0.1 (unmodified)\n```\n\n## Impact\n\nAn authenticated Tailscale user who should only have regular user access can:\n1. Bypass IP-based admin restrictions on the backend application by spoofing X-Forwarded-For to 127.0.0.1\n2. Appear as any arbitrary IP address in audit logs\n3. Bypass rate limiting tied to source IP\n4. Bypass geo-blocking or network-segment restrictions enforced by the backend\n\nThis is especially impactful because tsdproxy is designed as the sole access point for backend services that are otherwise network-isolated -- making the proxy the only enforcement boundary.\n\nFix: Add r.Out.Header.Del(\"X-Forwarded-For\") and r.Out.Header.Del(\"X-Real-IP\") in the Rewrite closure before calling r.SetXForwarded(). This ensures only the real Tailscale client IP appears in the XFF chain.\n\n### Credits\n\nReported by Vishal Shukla (@shukla304) using sechub.dev AI Agent\n\n### Support\n\nIf this disclosure work has been useful, [sponsoring](https://github.com/sponsors/therawdev) helps fund continued open-source security audits -- appreciated either way.",
"id": "GHSA-pqg7-v6wh-3pfp",
"modified": "2026-07-14T19:46:15Z",
"published": "2026-07-14T19:46:15Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/almeidapaulopt/tsdproxy/security/advisories/GHSA-pqg7-v6wh-3pfp"
},
{
"type": "WEB",
"url": "https://github.com/almeidapaulopt/tsdproxy/commit/e8200b7947719e5e7fbbbdb9c34f459a4c285e77"
},
{
"type": "PACKAGE",
"url": "https://github.com/almeidapaulopt/tsdproxy"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:L/I:H/A:N",
"type": "CVSS_V3"
}
],
"summary": "TsDProxy: X-Forwarded-For header injection allows IP spoofing in proxied requests to backend services"
}
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.