---
id: "CVE-2026-110278"
page: "https://infrastack.ai/live/CVE-2026-110278"
kind: "vulnerability"
aliases: ["CVE-2026-110278"]
title: "IBM ContextForge MCP Gateway ignores per-server identity propagation setting, sending user identity to upstream MCP servers"
severity: "high"
cvss: 7.4
epss: null
exploited: false
status: "open"
category: "ai-tooling"
products: [{"name": "IBM ContextForge MCP Gateway", "fixed": ["1.1.1"], "affected": ">= 1.0.0, <= 1.1.0", "ecosystem": "vendor"}]
ecosystems: ["vendor"]
tags: ["ai-tooling", "ibm-contextforge", "mcp", "self-hosted"]
not_affected_if: ["identity-propagation-disabled", "other-versions", "v1.1.1-or-later"]
published_at: "2026-11-13T22:17:20+00:00"
first_seen_at: "2026-11-13T22:33:33+00:00"
updated_at: "2026-11-13T22:35:25+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-110278", "first_seen_at": "2026-11-13T22:33:33+00:00"}, {"source": "analyst", "url": "https://www.ibm.com/support/pages/node/7296489", "first_seen_at": "2026-11-13T22:35:25+00:00", "cites": ["https://www.ibm.com/support/pages/node/7296489", "https://cdn.jsdelivr.net/gh/ibm/mcp-context-forge@main/docs/docs/manage/identity-propagation.md"]}]
exploit_sources: []
---

# IBM ContextForge MCP Gateway ignores per-server identity propagation setting, sending user identity to upstream MCP servers

Link for people: https://infrastack.ai/live/CVE-2026-110278 (this .md is for agents)

IBM ContextForge MCP Gateway versions 1.0.0 through 1.1.0 sign identity propagation claims with a static payload and skip per-gateway configuration checks during tool calls. An attacker can replay those signatures or obtain sensitive user identity information sent to third-party MCP servers.

_Summary, check and fix written by a model from the sources; labelled as such._

## Affected

- vendor:IBM ContextForge MCP Gateway affected >= 1.0.0, <= 1.1.0 fixed 1.1.1

## How to check

- Check the ContextForge MCP Gateway version: v1.0.0 through v1.1.0 is affected, v1.1.1 is fixed.
- Check whether identity propagation is on: IDENTITY_PROPAGATION_ENABLED=true in the gateway environment (default false), or a registered gateway with identity_propagation.enabled true. IBM says the per-server setting is ignored on tool calls and resource reads, so servers configured not to receive identity still get it. Also look at IDENTITY_SIGN_CLAIMS (HMAC-signed claims); IBM cites static signature payloads that can be replayed.

## Fix

Upgrade IBM ContextForge MCP Gateway to v1.1.1 (fixing PR https://github.com/IBM/mcp-context-forge/pull/7329). If you cannot upgrade, IBM says to set IDENTITY_PROPAGATION_ENABLED=false globally so no upstream server receives user identity, and restrict gateway access to trusted users.

## Timeline

- 2026-11-13T22:33:33+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-110278
- 2026-11-13T22:35:25+00:00 analyst https://www.ibm.com/support/pages/node/7296489 — Read IBM bulletin (node/7296489): affected v1.0.0-v1.1.0, fixed v1.1.1, workaround IDENTITY_PROPAGATION_ENABLED=false. Not exploited, no PoC seen. Skipped social search (obscure).

## Documents

- [advisory] IBM Security Bulletin: ContextForge MCP Gateway information exposure https://www.ibm.com/support/pages/node/7296489
- [describes] ContextForge Identity Propagation docs https://cdn.jsdelivr.net/gh/ibm/mcp-context-forge@main/docs/docs/manage/identity-propagation.md

## From the sources

IBM ContextForge MCP Gateway 1.0.0 through 1.1.0 could allow an attacker to replay identity propagation claims signatures or expose sensitive user identity information to third-party MCP servers due to static signature payloads and missing per-gateway configuration checks during tool invocation.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_