GHSA-g2qj-prgh-4g9r
HIGHGHSA-g2qj-prgh-4g9r is a high-severity (CVSS 7.5) Information Exposure vulnerability in github.com/nhost/nhost. O3 Security confirms whether GHSA-g2qj-prgh-4g9r is actually reachable in your code before you act, and blocks exploitation at runtime until you patch.
Nhost Leaks Refresh Tokens via URL Query Parameter in OAuth Provider Callback
Blast Radius
github.com/nhost/nhostReal-time download stats are indexed for npm and PyPI packages. This vulnerability affects Go packages — download data is not available via public APIs for these ecosystems.
Description
Refresh Token Leaked via URL Query Parameter in OAuth Provider Callback
Summary
The auth service's OAuth provider callback flow places the refresh token directly into the redirect URL as a query parameter. Refresh tokens in URLs are logged in browser history, server access logs, HTTP Referer headers, and proxy/CDN logs.
Note that the refresh token is one-time use and all of these leak vectors are on owned infrastructure or services integrated by the application developer.
Affected Component
- Repository:
github.com/nhost/nhost - Service:
services/auth - File:
services/auth/go/controller/sign_in_provider_callback_get.go - Function:
signinProviderProviderCallback(lines 257-261)
Root Cause
In sign_in_provider_callback_get.go:257-261, after successful OAuth sign-in, the refresh token is appended as a URL query parameter:
if session != nil {
values := redirectTo.Query()
values.Add("refreshToken", session.RefreshToken)
redirectTo.RawQuery = values.Encode()
}
This results in a redirect like:
HTTP/1.1 302 Found
Location: https://myapp.com/callback?refreshToken=a1b2c3d4-e5f6-7890-abcd-ef1234567890
Proof of Concept
Step 1: Initiate OAuth login
GET /signin/provider/github?redirectTo=https://myapp.com/callback
Step 2: Complete OAuth flow with provider
Step 3: Auth service redirects with token in URL
HTTP/1.1 302 Found
Location: https://myapp.com/callback?refreshToken=a1b2c3d4-e5f6-7890-abcd-ef1234567890
Step 4: Token is now visible in owned infrastructure and services:
Browser History:
# User's browser history now contains the refresh token
HTTP Referer Header:
# If the callback page loads ANY external resource (image, script, etc.):
GET /resource.js HTTP/1.1
Host: cdn.example.com
Referer: https://myapp.com/callback?refreshToken=a1b2c3d4-e5f6-...
# Note: modern browsers default to strict-origin-when-cross-origin policy,
# which strips query parameters from cross-origin Referer headers.
# Additionally, the Referer is only sent to services integrated by the
# application developer (analytics, CDNs, etc.), not arbitrary third parties.
Server Access Logs:
# Reverse proxy, CDN, or load balancer logs on owned infrastructure:
2026-03-08 12:00:00 GET /callback?refreshToken=a1b2c3d4-e5f6-... 200
Step 5: Attacker uses stolen refresh token
# Exchange stolen refresh token for new access token
curl -X POST https://auth.nhost.run/v1/token \
-H 'Content-Type: application/json' \
-d '{"refreshToken": "a1b2c3d4-e5f6-7890-abcd-ef1234567890"}'
# Note: refresh tokens are one-time use, so this only works if the
# legitimate client has not already consumed the token and if the attacker has
# compromised your infrastructure to get access to this information
Impact
-
Session Hijacking: Anyone who obtains the token before it is consumed by the legitimate client can generate new access tokens, though the refresh token is one-time use and cannot be reused after consumption.
-
Leak Vectors: URL query parameters are visible in owned infrastructure and integrated services:
- Browser history (local access)
- HTTP Referer headers (mitigated by modern browser default referrer policies; only sent to developer-integrated services)
- Server access logs (owned infrastructure)
- Proxy/CDN/WAF logs (owned infrastructure)
-
Affects All OAuth Providers: Every OAuth provider flow (GitHub, Google, Apple, etc.) goes through the same callback handler.
Fix
Implemented PKCE (Proof Key for Code Exchange) for the OAuth flow. With PKCE, the authorization code cannot be exchanged without the code_verifier that only the original client possesses, preventing token misuse even if the URL is logged.
See: https://docs.nhost.io/products/auth/pkce/
Resources
- OWASP: Session Management - Token Transport: "Session tokens should not be transported in the URL"
- RFC 6749 Section 10.3: "Access tokens and refresh tokens MUST NOT be included in the redirect URI"
- CWE-598: Use of GET Request Method With Sensitive Query Strings
- CWE-200: Exposure of Sensitive Information to an Unauthorized Actor
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 🐹Go | github.com/nhost/nhost | all versions | 0.0.0-20260330133707-294954e0fc3a |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for github.com/nhost/nhost. O3's reachability analysis confirms whether the vulnerable code path is actually invoked in your application, so you act on real exposure instead of every transitive match.
Fix
Update github.com/nhost/nhost to 0.0.0-20260330133707-294954e0fc3a or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms GHSA-g2qj-prgh-4g9r is resolved across your whole dependency graph.
Workarounds
If you can't upgrade right away: gate or disable the affected feature, validate untrusted input at the boundary, and avoid passing attacker-controlled data into the vulnerable path. O3's runtime protection blocks exploitation in production as an interim safeguard until the upgrade lands.
How O3 protects you
O3 pinpoints whether GHSA-g2qj-prgh-4g9r is reachable in your code and exactly where to fix it, then blocks exploitation in production at runtime until the patched version is deployed.
Tailored to GHSA-g2qj-prgh-4g9r. Runtime protection reduces exposure until a permanent patch is applied and verified — it complements patching, it doesn't replace it.
Frequently Asked Questions
Is GHSA-g2qj-prgh-4g9r in your dependencies?
O3 detects GHSA-g2qj-prgh-4g9r across Go dependencies and uses function-level reachability to confirm whether the vulnerable code path is actually reachable — not just present. No false positives.