/

CVE-2026-59973

Severity:

HIGH RISK

(8.5)

Server-Side Request Forgery (CWE-918) in FrontMCP Adapter

CVE-2026-59973

Deferred

Vulnerability type

Category:

Server-Side Request Forgery

CWE:

CWE-918

Impact:

Internal Network Access and Sensitive Data Exposure

Severity:

HIGH RISK

(8.5)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A server-side request forgery (SSRF) vulnerability in the FrontMCP OpenAPI adapter allows an authenticated user who can configure or import an OpenAPI specification to influence backend-origin requests.

In affected deployments, hostname resolution, redirects, or IPv4-mapped IPv6 representations may bypass external-reference checks and allow requests to internal services, including administrative or metadata-like endpoints.

Affected context

The issue occurs in FrontMCP and its OpenAPI-related adapters when untrusted OpenAPI URLs, specifications, or external reference-resolution options are passed to OpenAPI tooling. Risk is greater in hosted or multi-user deployments where authenticated users can configure or import specifications and the backend can reach private network destinations. Deployments limited to trusted local administrators, or those that disable external reference protocols, may have reduced exposure.

Exploit preconditions

  • Attack surface: OpenAPI specification import and external $ref resolution performed by the backend
  • Required access: Authenticated ability to configure or import an OpenAPI specification in the deployment
  • Trigger point: Loading an OpenAPI document or resolving an external reference
  • Why exploitation is possible: The external-reference guard validates hostname strings without consistently resolving and pinning destination addresses, rechecking redirect targets, or normalizing IPv4-mapped IPv6 addresses

Why this happens

The underlying bug is insufficient validation of server-side request destinations, mapping to CWE-918: Server-Side Request Forgery. The adapter relies on hostname-based checks while allowing a downstream HTTP client to resolve names, follow redirects, and interpret alternate IP address forms. Because validation is not performed against the final, resolved destination, attacker-influenced references may reach private or loopback services.

Potential impact

  • Backend requests could reach internal administrative APIs
  • Cloud or infrastructure metadata-like services may be exposed
  • Sensitive responses from private network endpoints could be returned or processed
  • Internal service availability or configuration could be affected
  • The application may become a pivot point for further network exploration

General mitigation guidance

Mitigation generally focuses on enforcing destination-aware egress controls and reducing trust in user-supplied OpenAPI references.

Common defensive patterns include:

  • Disable external reference protocols when they are not required
  • Resolve hostnames and validate the resulting addresses before connecting
  • Pin validated destination addresses and prevent resolution changes during the request
  • Revalidate every redirect target and reject private, loopback, link-local, multicast, and reserved ranges
  • Normalize and consistently validate IPv4, IPv6, and IPv4-mapped IPv6 representations
  • Apply network egress policies that restrict backend access to approved destinations
  • Run request-processing components with minimal network and service privileges
  • Prefer trusted, preapproved OpenAPI sources in multi-user environments

How teams detect issues like this

Organizations can identify SSRF risks through:

  • Static analysis for user-controlled URLs passed to HTTP clients or OpenAPI loaders
  • Code review checks for hostname-only validation, unrestricted redirects, and DNS rebinding exposure
  • Tests covering loopback, private ranges, link-local addresses, IPv4-mapped IPv6, redirects, and changing DNS responses
  • Network monitoring for backend requests to internal address ranges or metadata-like endpoints
  • Egress policies and alerts that flag unexpected connections from application workloads
  • Automated security review tools that model URL flow and server-side request behavior

Sources

SourceLink
github.comgithub.com/agentfront/frontmcp/commit/96a78eaa5c6c4bc51cced557d83d1a03344cb03d
github.comgithub.com/agentfront/frontmcp/pull/496
github.comgithub.com/agentfront/frontmcp/releases/tag/v1.5.0
github.comgithub.com/agentfront/frontmcp/security/advisories/GHSA-65h7-9wrw-629c
github.comgithub.com/agentfront/mcp-from-openapi/commit/be3409cce6e97642696d4ee5a4e4e2712490b277
github.comgithub.com/agentfront/mcp-from-openapi/pull/7
github.comgithub.com/agentfront/mcp-from-openapi/releases/tag/v2.5.0

Frequently asked

questions

Exploitation generally requires authenticated access to import or configure an OpenAPI specification. The exact scope depends on how the deployment exposes configuration features.

In those environments, less-trusted users may be able to influence specifications while the backend retains access to private network services that users cannot reach directly.

Disable external reference protocols unless necessary, and enforce destination-aware validation that covers DNS resolution, redirects, address normalization, and private-network ranges.

Combine strict URL and address validation with allowlisted egress, redirect controls, network segmentation, least privilege, and monitoring for unexpected backend-origin connections.

Find out what's already
exploitable in your codebase.