概要
Identity providers do the heavy lifting — auth flows, MFA, social login, account merging, JWT issuance — and they do it well. The defaults are tuned for time-to-launch rather than security, which is reasonable for a developer tool but matters once your app sees production traffic. Clerk and Auth0 are the modern leaders, both ship a public configuration discovery surface that's perfectly fine in isolation and quietly informative when combined with other recon. Permissive redirect URI allowlists, exposed tenant discovery, social-provider lists that reveal your employee SSO setup — each is a small leak; together they help target the social engineering that follows.
仕組み
Both providers expose tenant configuration via discovery endpoints. Clerk's `/v1/environment` returns the tenant's full client-side config: enabled auth methods, social providers (Google, GitHub, etc.), allowed redirect URLs, organization features, custom branding. Auth0's `/.well-known/openid-configuration` and `/.well-known/jwks.json` are OIDC-standard discovery — required for the protocol — but they reveal the issuer URL, supported algorithms, and the tenant's public key. Beyond discovery, common configuration mistakes include overly broad `redirect_uri` allowlists (wildcards or partial matches), unused social providers left enabled, and JWT signing keys with weak rotation policies.
バリエーション
Permissive redirect URI
Allowlist accepts wildcards or partial matches (`https://*.yourdomain.com/*`). Attacker registers a subdomain that matches and harvests authorization codes through their own callback.
Unused social provider enabled
GitHub login enabled but nobody uses it; remains a vector for account hijack via social-account takeover (attacker compromises victim's GitHub, signs in to your app via SSO).
Tenant discovery for spear-phishing setup
Public discovery endpoint reveals branding, organization name, configured providers. Attacker crafts a phishing page that exactly mirrors the real SSO flow.
Long-lived JWT signing keys
Without rotation, a single JWKS compromise (insider, supply chain) compromises every token issued for the lifetime of the key.
被害範囲
Social engineering leverage dominates: knowing every redirect URI helps craft pixel-perfect phishing pages; knowing the tenant's IdP list helps spear-phish employees through familiar-looking SSO flows. Direct impact when redirect URI allowlists are too permissive — attacker harvests authorization codes via a controlled subdomain. Account takeover via unused social providers when an employee's external account gets compromised. JWT-issuer compromise is rare but devastating: every token signed with the leaked key is forgeable until rotation.
// what fixvibe checks
What FixVibe checks
FixVibe checks backend-as-a-service exposure through safe configuration and access-boundary signals. Reports focus on what is exposed and how to lock it down. For check-specific questions about exact detection heuristics, active payload details, or source-code rule patterns, contact support@fixvibe.app.
鉄壁の防御
Restrict allowed redirect URIs to exact matches. Audit your Auth0 application's 'Allowed Callback URLs' and 'Allowed Logout URLs' lists; the same for Clerk's redirect URL configuration. Disable social providers your users aren't actually using — every enabled provider is an additional account-takeover surface. Rotate JWT signing keys on a schedule (annual is fine for low-risk apps; quarterly for high-risk); both Clerk and Auth0 support overlap-windows so existing tokens remain valid during rotation. For Auth0, configure tenant-level rate limits on auth endpoints. For Clerk, audit the publishable-key bindings and tenant settings — the dashboard surfaces drift from recommended baseline. As a defense-in-depth layer, monitor for anomalous auth events (mass signups, cross-region logins) — both providers offer this telemetry.