CVE-2026-64684
Medium · CVSS 6.8Published 2026-09-16
CVSS
6.8
Medium
EPSS
—
exploit probability
Active Exploit
No
not yet listed
What is CVE-2026-64684?
RMCP is an official Rust SDK for the Model Context Protocol. Prior to 2.1.0, the rmcp crate's StreamableHttpClientTransport in crates/rmcp/src/transport/common/reqwest/streamable_http_client.rs builds its default_http_client with reqwest's automatic redirect policy and applies caller-supplied values from StreamableHttpClientTransportConfig.custom_headers without marking them as sensitive. When a malicious or compromised MCP endpoint returns a cross-origin 307 or 308 redirect, reqwest follows the redirect and apply_custom_headers causes custom API keys or authentication tokens to be replayed to the new origin, where an attacker can capture and reuse them. The separate auth_header path is not affected because it uses the standard Authorization header, which reqwest strips on cross-origin redirects. This issue is fixed in version 2.1.0.
CVSS Score Explained
CVSS 6.8 is Medium severity. Exploitation usually requires some specific conditions (local access, user interaction, or a particular config). Still worth patching in your next maintenance window.
EPSS Exploitation Probability
No EPSS score yet — model needs more data on this CVE.
References & Patches
- ›https://github.com/modelcontextprotocol/rust-sdk/commit/496902b9cf2c8a947454718da31829ae776b969b
- ›https://github.com/modelcontextprotocol/rust-sdk/pull/936
- PATCHhttps://github.com/modelcontextprotocol/rust-sdk/releases/tag/rmcp-v2.1.0
- ›https://github.com/modelcontextprotocol/rust-sdk/security/advisories/GHSA-9g45-5xwm-f3wc
Track vulnerabilities for your stack
Tell us what tech you run — Node.js, Python, Java, nginx, whatever — and get a weekly email when new CVEs affect it.
Recent CVEs
- CVE-2026-93435redis-parser through 3.0.0 contains a denial of service vulnerability in the RESP protocol parser that allows malicious Redis endpoints to crash the client process through unbounded recursion on nested arrays. Attackers can send crafted RESP byte streams with repeated array headers that exhaust the V8 call stack, causing an uncaught RangeError that terminates the Node.js process without triggering error handling callbacks.8.7
- CVE-2026-93436vLLM through 0.29.0 fails to properly clean up decode-side metadata for rejected inference requests in prefill/decode disaggregated deployments. Remote attackers can submit requests with max_tokens=0 to exhaust decode-worker memory without bound until the worker restarts.8.7
- CVE-2026-85885Improper neutralization of special elements used in a command ('command injection') in M365 Copilot allows an authorized attacker to elevate privileges over a network.9.9
- CVE-2026-85889Missing authentication for critical function in Azure AI Foundry allows an unauthorized attacker to elevate privileges over a network.10.0
- CVE-2026-85917Server-side request forgery (ssrf) in Azure AI Foundry allows an unauthorized attacker to elevate privileges over a network.7.5
- CVE-2026-87701Improper neutralization of special elements in output used by a downstream component ('injection') in Azure Cosmos DB allows an authorized attacker to elevate privileges over a network.9.6