Threat
When the Observability Layer Becomes the Attack Surface — Splunk’s MCP Server and the Protocol Vulnerability Nobody Patched
First reported · Discovered forkast.news
Page published
Earliest dated coverage: 8 Oct 2026 · First observed: 15 Oct 2026 · Latest dated coverage: 8 Oct 2026
Coverage timeline
Single-source analysis — one report is available.
Why it matters
Splunk's MCP Server places AI agents directly on top of enterprise logs, alerts, and search, so a specification-level MCP data-leak flaw combined with a critical RCE turns the observability layer into a high-value attack surface for agentic AI deployments.
Forkast reports that Splunk's MCP Server, which gives AI agents protocol-level access to enterprise observability data, is exposed to two compounding risks: a structural flaw in the Model Context Protocol specification itself (disclosed by ClawSecure, where MCP servers automatically fetch attacker-controlled links enabling data leakage without model interaction) and CVE-2026-76404 (CVSS 9.1), a Remote Code Execution flaw from insecure deserialization in the server's credential management, patched in version 1.2.1 under advisory SVD-2026-0808. ClawSecure tested 14 models across five labs and found all failed to consistently block the protocol-level threats.