Veille CVE

Recherche « agentfront frontmcp »

2 CVE

Critique 1 Élevée 1 Moyenne 0 Faible 0 Non évaluée 0 Exploitées (KEV) Tri : plus récentes « agentfront frontmcp » ✕
  1. CVE-2026-59973 8.5 Élevée

    agentfront frontmcp · agentfront mcp-from-openapi · @frontmcp adapters

    FrontMCP is a TypeScript-first framework for the Model Context Protocol (MCP). From mcp-from-openapi 2.3.0 until 2.5.0 and from frontmcp and @frontmcp/adapters 1.2.1 until 1.5.0, libs/adapters/src/openapi/openapi.adapter.ts loadOpenAPISpec() forwards untrusted OpenAPI url and spec inputs and loadOptions.refResolution to OpenAPIToolGenerator.fromURL() and OpenAPIToolGenerator.fromJSON(). The external $ref guard checks parsed hostname strings without resolving addresses, pinning validated addresses, revalidating redirect targets, or normalizing IPv4-mapped IPv6. An authenticated user who can import or configure an OpenAPI specification in a hosted or multi-user deployment can use DNS-to-loopback resolution, redirect-to-loopback behavior, or IPv4-mapped IPv6 loopback forms to cause backend-origin requests to internal services. This can expose internal administrative APIs, metadata-like services, and other private network endpoints. The practical impact is lower when only a trusted local administrator can configure OpenAPI specs, and disabling external reference protocols prevents the external $ref request. This issue is fixed in mcp-from-openapi 2.5.0 and frontmcp and @frontmcp/adapters 1.5.0.

  2. CVE-2026-67531 9.3 Critique

    agentfront frontmcp

    FrontMCP is a TypeScript-first framework for the Model Context Protocol (MCP). Prior to 1.5.7, the sandboxed codecall:execute tool exposes live host Zod schema instances to the script via getTool(), and because Zod v4 defines _zod as a non-configurable, non-writable own property, the ECMAScript Proxy invariants force the security membrane to hand back the raw host object, letting a script reach _zod.constr.constructor (the host Function constructor) and execute arbitrary code in the server process. A single tools/call is sufficient to escape the sandbox and achieve remote code execution as the server user, exposing everything the process holds such as OAuth client secrets, JWT_SECRET, session keys, database credentials, and cloud instance metadata. Because the framework's DEFAULT_AUTH_OPTIONS is public mode, an unconfigured server serves this to unauthenticated callers, and on authenticated servers an indirect prompt injection in tool output or fetched content can trigger it without a human attackerThis issue is fixed in version 1.5.7.