# CVE-2026-123272: Mark3labs mcp-go improper validation in OAuthHandler

Source: https://junglewise.ai/threats/cve-2026-123272-mark3labs-mcp-go-before-v1-2-1-contains-an-improper-validation  
Publisher: Junglewise Threat Intelligence, https://junglewise.ai/threats  
Published: 2026-11-05. Updated: 2026-11-05  
License: CC BY 4.0 with a link back

CVE-2026-123272 · Severity: high · CVSS 8 · Published 2026-11-05

Mark3labs mcp-go is a Go implementation of the Model Context Protocol (MCP), used for secure communication between applications and language model servers. A validation flaw in the OAuth handler allows attackers to trick the application into obtaining tokens for arbitrary resources by crafting malicious MCP server metadata, enabling token theft and replay attacks against unintended services.

Technologies: [Mark3labs Mcp-Go](https://junglewise.ai/threats/technologies/mcp-go). Vendors: [Mark3labs](https://junglewise.ai/threats/vendors/mark3labs).

## Quick answers

**Is CVE-2026-123272 exploited in the wild?** No. Junglewise Threat Intelligence has no report of CVE-2026-123272 being exploited in the wild.

**How severe is CVE-2026-123272?** High severity, with a CVSS score of 8, published 5 November 2026.

**Which products and versions are affected?** Mark3labs Mcp-Go before v1.2.1.

**Is there a fix for CVE-2026-123272?** Yes: fixed in v1.2.1. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.

**How is CVE-2026-123272 attacked?** It is attacked over the network.

**Am I affected by CVE-2026-123272?** Check whether you run Mark3labs Mcp-Go at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

## How to fix

- Fixed in v1.2.1

## Technical details

The OAuthHandler.fetchServerMetadata function fails to validate that resources declared in well-known OAuth metadata actually bind to the addressed MCP server, violating RFC 9728 section 3.3. An attacker running a malicious MCP server can declare an arbitrary resource as the RFC 8707 resource indicator, causing the client's authorization server to mint tokens that the attacker can replay against the declared resource.

## Affected products

- Mark3labs mcp-go before v1.2.1

## Timeline

- 2026-11-05: disclosed

## References

- [https://github.com/mark3labs/mcp-go/blob/4cb0d2015f24b669726fd8e5e53fc8c6d3f57abf/client/transport/oauth.go](https://github.com/mark3labs/mcp-go/blob/4cb0d2015f24b669726fd8e5e53fc8c6d3f57abf/client/transport/oauth.go)
- [https://github.com/mark3labs/mcp-go/commit/90d247ed8d45cbb7d71d9a95e112789b78760443](https://github.com/mark3labs/mcp-go/commit/90d247ed8d45cbb7d71d9a95e112789b78760443)

## Related threats

- [mark3labs mcp-go SSRF and credential leak via unvalidated Location header](https://junglewise.ai/threats/go-2026-6911-ssrf-and-credential-leak-via-unvalidated-location-header-1f243351)
- [mcp-go denial of service in StreamableHTTPServer.ServeHTTP](https://junglewise.ai/threats/cve-2026-108859-mcp-go-through-1-2-1-contains-a-denial-of-service-vulnerability)
- [mcp-go HTTP Host header validation bypass](https://junglewise.ai/threats/cve-2026-81092-mcp-go-http-host-header-validation-bypass)