• Injection

SSRF (Server-Side Request Forgery)

SSRF tricks your server into making HTTP (or related) requests to destinations you never intended — often internal cloud metadata, admin panels, or private network hosts.

How attackers abuse it

Features that fetch URLs on behalf of users — link previews, webhooks, PDF generators, avatar importers, CI callbacks — are classic SSRF entry points. An attacker supplies http://169.254.169.254/ or an internal hostname and the server reaches infrastructure the attacker cannot hit directly.

Blind SSRF may only change timing or error text; full-read SSRF returns response bodies. Attackers chain redirects, DNS rebinding, and alternate schemes (file://, gopher:// where supported) to bypass naive blocklists.

In cloud environments, metadata endpoints can leak temporary credentials. That turns a “fetch URL” bug into account takeover of cloud resources.

What proof looks like

Proof shows the application initiating a request to an attacker-controlled or in-scope internal target and returning evidence of that fetch (body snippet, timing channel under rules of engagement, or blocked-but-attempted destination in logs). Safe PoCs avoid destructive payloads and stay within authorized scope.

In PandaONE, this class maps to the Injection finding category. Alerts ship only with evidence — never as unverified scanner noise.

How to prevent it

  • Prefer allowlists of destinations over denylists of private IPs.
  • Resolve DNS yourself, block link-local / private / metadata ranges, and disable following redirects to untrusted hosts.
  • Use dedicated egress proxies with hard network policy for server-side fetches.
  • Never expose raw response bodies from arbitrary URLs to end users without sanitization.
  • Separate “user-supplied URL” features from privileged network paths.

• FAQ

Frequently asked questions

← All vulnerability classes

Hunt this class with proof

Or scan your domain.

Featured on ScrollLaunchFeatured on CodeHype