Summary
Arcane, a Docker management interface, contained an unauthenticated server-side request forgery (SSRF) issue in the /api/templates/fetch endpoint prior to version 1.17.3.
By supplying an arbitrary URL, an attacker could coerce the Arcane server into making HTTP requests to internal or external resources and relay the responses back, potentially exposing internal services or metadata.
Affected context
The vulnerability occurs in Arcane instances that expose the /api/templates/fetch endpoint over a network, particularly when the service is reachable from the public internet.
The endpoint trusts a user-supplied url parameter and uses it as the target for an HTTP GET request without authenticating the caller or validating the scheme or host.
This becomes especially risky when Arcane can reach internal networks, cloud metadata endpoints, or other protected services that are not directly reachable by the attacker.
Exploit preconditions
- Attack surface: HTTP(S) access to the Arcane /api/templates/fetch endpoint.
- Required access: None; the vulnerability is exploitable without authentication.
- Trigger point: Supplying a crafted
urlparameter that points to attacker-chosen internal or external addresses. - Why exploitation is possible: The backend performs a server-side HTTP GET using the supplied URL with no scheme or host validation and returns the raw response to the caller, effectively exposing the server's network position as a proxy.
Why this happens
The root cause aligns with CWE-918: Server-Side Request Forgery (SSRF).
The application directly interpolates a user-controlled url parameter into a backend HTTP client call without enforcing a strict allowlist, validating destination hosts, or constraining protocols.
Because there is no authentication on the endpoint, any external party can leverage the server's network privileges to initiate requests to internal services, metadata endpoints, or other resources that are not intended to be exposed.
Potential impact
- Disclosure of information from internal HTTP(S) services that are otherwise not publicly exposed.
- Access to cloud or container metadata services that may reveal sensitive credentials or configuration.
- Use of Arcane as a pivot point for scanning internal networks and enumerating services.
- Potential interaction with internal APIs in a way that could change state if those APIs trust requests from Arcane's host or network.
- Abuse of Arcane's outbound network connectivity to contact external targets, possibly aiding in anonymizing malicious traffic.
General mitigation guidance
Effective mitigation generally focuses on constraining where the server can connect and how user input influences outbound requests:
- Enforce strict allowlists for outbound URLs, limiting requests to trusted domains or specific templates repositories.
- Reject or sanitize user-provided URLs that reference private address ranges, loopback interfaces, link-local addresses, or cloud metadata endpoints.
- Restrict accepted URL schemes to safe protocols (e.g., HTTPS only) and disallow potentially dangerous or unnecessary schemes.
- Apply authentication and authorization controls to endpoints that perform server-side network requests, ensuring only trusted users can invoke them.
- Use network-level controls (firewalls, container network policies) to restrict the server's ability to reach sensitive internal services.
- Adopt secure-by-default patterns where templates and resources are selected from predefined configurations instead of arbitrary user-specified URLs.
How teams detect issues like this
Detection can focus on both code-level and operational signals:
- Static analysis: Search for HTTP client calls that use user-controlled URL parameters without validation or allowlisting.
- Code review: Flag endpoints that proxy arbitrary URLs, especially when unauthenticated or exposed over the internet, and verify the presence of scheme/host validation.
- Unit and integration tests: Include security-focused tests that attempt requests to private IP ranges, localhost, and metadata endpoints, ensuring they are rejected.
- Runtime monitoring: Inspect access logs for unusual patterns in the
urlparameter, such as internal IP ranges (e.g., 10.x, 172.16–31.x, 192.168.x), 127.0.0.1, or metadata service hostnames. - Network monitoring: Use IDS/IPS or network telemetry to detect unexpected outbound requests from the Arcane host/container to internal segments or cloud metadata addresses.

