CVEReports
CVEReports

Automated vulnerability intelligence platform. Comprehensive reports for high-severity CVEs generated by AI.

Product

  • Home
  • Sitemap
  • RSS Feed

Company

  • About
  • Contact
  • Privacy Policy
  • Terms of Service

© 2026 CVEReports. All rights reserved.

Made with love by Amit Schendel & Alon Barad



CVE-2026-105806

CVE-2026-105806: Privilege Escalation and Missing Authorization in @payloadcms/plugin-mcp

Alon Barad
Alon Barad
Software Engineer

Oct 7, 2026·5 min read·22 visits

Executive Summary (TL;DR)

A missing authorization check in @payloadcms/plugin-mcp allows authenticated low-privilege users to generate MCP API keys mapped to administrative accounts, resulting in privilege escalation and full account takeover.

CVE-2026-105806 is an improper access control vulnerability within the Model Context Protocol (MCP) plugin for Payload CMS. Authenticated users with low privileges can manipulate API key creation and mapping to associate keys with arbitrary users, including administrators. This allows total session takeovers and privilege escalation via MCP-authenticated API requests.

Vulnerability Overview

The Model Context Protocol (MCP) plugin for Payload CMS integrates LLM-based agent capabilities with the database layer of Payload CMS. It exposes a specific collection, payload-mcp-api-keys, which manages authentication tokens for MCP clients.\n\nUnder normal operations, these API keys associate client sessions with specific CMS user identities, granting them the capability to run automated actions inside Payload CMS. The security model relies on these keys executing precisely within the permission scope of their linked user.\n\nHowever, prior to version 3.88.0, the collection defining these keys omitted collection-level and field-level access controls. This omission allowed any authenticated user to interact with the database records of other users or generate new keys mapped directly to high-privilege administrators.

Root Cause Analysis

The root cause of CVE-2026-105806 is the absence of authorization checks on the payload-mcp-api-keys collection and its associated relationship fields.\n\nIn Payload CMS, database collections without explicit access properties default to allowing read, write, and update permissions for any authenticated user. Because the plugin defined the payload-mcp-api-keys collection without restriction, any authenticated entity could submit requests to retrieve or alter API keys.\n\nFurthermore, the user relationship field lacked restrictions, allowing client-supplied input to directly map the owner ID during creation or modification operations. This combination allowed an attacker to bypass user boundaries, associating their generated API key with an administrative identifier.

Code Analysis

Prior to the patch, the payload-mcp-api-keys collection did not implement restrictive access control hooks. The user field allowed direct user-supplied mutations.\n\nmermaid\ngraph LR\n A["Attacker (Low Privilege)"] -->|1. POST /api/payload-mcp-api-keys| B["Payload CMS API"]\n B -->|2. Stores user: admin-id| C[("Database")]\n C -->|3. Issues Administrative Token| A\n\n\nThe patch in commit 025581d8c5ed836039ad85397288a30f0d3c11c1 mitigated this by adding strict validation checks. Let us examine the modifications introduced in packages/plugin-mcp/src/collections/createApiKeysCollection.ts:\n\nts\n const isAuthenticated = ({ req }: { req: PayloadRequest }): boolean =>\n Boolean(req.user && req.user.collection === userCollection)\n\n const restrictToOwnKeys = ({ req }: { req: PayloadRequest }): boolean | Where =>\n req.user && req.user.collection === userCollection ? { user: { equals: req.user.id } } : false\n\n return {\n slug: 'payload-mcp-api-keys',\n access: {\n create: isAuthenticated,\n delete: restrictToOwnKeys,\n read: restrictToOwnKeys,\n unlock: restrictToOwnKeys,\n update: restrictToOwnKeys,\n },\n\n\nAdditionally, the user field access controls were set to false for direct creation and updates from client inputs, forcing the use of a server-side default value hook:\n\nts\n {\n name: 'user',\n type: 'relationship',\n access: {\n create: () => false,\n update: () => false,\n },\n defaultValue: ({ req }: { req: PayloadRequest }) =>\n req.user && req.user.collection === userCollection ? req.user.id : undefined,\n relationTo: userCollection as CollectionSlug,\n required: true,\n },\n\n\nThis architecture guarantees that the database always resolves the API key's owner directly from the verified server-side session variable (req.user.id), eliminating the potential for identity spoofing.

Exploitation Methodology

An attacker must first authenticate to Payload CMS using a legitimate, low-privilege account. No special elevated roles or permissions are required to launch this attack.\n\nThe attacker then requests a list of administrators or obtains an administrator ID via standard enumeration methods, which is a common pattern in multi-user content management systems.\n\nNext, the attacker submits an API key creation payload to the MCP endpoint. By explicitly passing the targeted administrative user's ID within the user relationship field, the attacker registers a new API key.\n\nWhen the system processes subsequent requests presenting this API key, the authentication layer maps the session to the target administrative user. The attacker can then perform administrative actions and execute system tools with full privileges.

Impact Assessment

The impact of CVE-2026-105806 is rated as High, with a CVSS v4.0 score of 8.6. Exploitation allows complete privilege escalation and full account takeover.\n\nBecause the MCP plugin interacts directly with the database layer, an administrative token gives the attacker full write and read capabilities across all collections. This includes accessing sensitive proprietary data, user credentials, and configuration files.\n\nIf the CMS manages critical integration systems or runs system-level tasks, the attacker could leverage the MCP engine to execute arbitrary code. The lack of operational complexity makes this a severe risk for enterprise installations running vulnerable versions of the plugin.

Remediation and Mitigation

The primary remediation step is to upgrade @payloadcms/plugin-mcp to version 3.88.0 or higher, which secures the default collection properties.\n\nIf upgrading is not immediately possible, administrators must restrict access to the payload-mcp-api-keys collection manually or disable the MCP plugin entirely. Deploying custom access control hooks on the collection schema can temporarily mitigate the vulnerability.\n\nFor custom workflows that require cross-user management of API keys by administrators, developers must use the overrideApiKeyCollection hook to safely enforce access control logic. Below is a secure configuration demonstrating how to limit management privileges strictly to administrative users:\n\nts\nimport type { PayloadRequest } from 'payload'\nimport { mcpPlugin } from '@payloadcms/plugin-mcp'\n\nconst isAdmin = ({ req }: { req: PayloadRequest }) =>\n Boolean(\n req.user &&\n req.user.collection === 'users' &&\n 'role' in req.user &&\n req.user.role === 'admin',\n )\n\nmcpPlugin({\n overrideApiKeyCollection: (collection) => {\n collection.access = {\n create: isAdmin,\n delete: isAdmin,\n read: isAdmin,\n unlock: isAdmin,\n update: isAdmin,\n }\n collection.fields = collection.fields.map((field) =>\n 'name' in field && field.name === 'user'\n ? { ...field, access: { create: isAdmin, update: isAdmin } }\n : field,\n )\n return collection\n },\n})\n

Official Patches

Payload CMSOfficial Patch Commit

Fix Analysis (1)

Technical Appendix

CVSS Score
8.6/ 10
CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:N/SC:N/SI:N/SA:N

Affected Systems

Payload CMS@payloadcms/plugin-mcp

Affected Versions Detail

Product
Affected Versions
Fixed Version
@payloadcms/plugin-mcp
Payload CMS
>= 3.61.0, < 3.88.03.88.0
AttributeDetail
CWE IDCWE-862
Attack VectorNetwork
CVSS v4.08.6
Exploit StatusUnproven / None
KEV StatusNot Listed

MITRE ATT&CK Mapping

T1068Exploitation for Privilege Escalation
Privilege Escalation
CWE-862
Missing Authorization

The software does not perform an authorization check when an actor attempts to access a resource or perform an action.

Vulnerability Timeline

Security patch commit authored
2026-08-11
CVE-2026-105806 published to NVD
2026-10-06
GitHub Advisory GHSA-2q76-m6w6-qgc6 published
2026-10-06

References & Sources

  • [1]GitHub Security Advisory GHSA-2q76-m6w6-qgc6
  • [2]Official Patch Commit
  • [3]Release v3.88.0
  • [4]CVE-2026-105806 Record

Attack Flow Diagram

Press enter or space to select a node. You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.
Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

More Reports

•about 2 hours ago•CVE-2026-58254
6.5

CVE-2026-58254: Authorization Bypass in NATS Server Message Tracing for Leaf Node Connections

NATS Server prior to versions 2.12.8 and 2.14.3 contains an incorrect authorization vulnerability (CWE-863) in its message tracing subsystem. When evaluating whether a connection is permitted to publish trace events to a destination subject specified in the Nats-Trace-Dest header, the server only checks standard client permissions and omits permission checks for leaf node connections. This allows a connected leaf node operator with restricted publishing privileges to emit internal trace events to unauthorized subjects and suppress standard message delivery using the Nats-Trace-Only header.

Alon Barad
Alon Barad
3 views•10 min read
•about 3 hours ago•CVE-2026-63341
7.5

CVE-2026-63341: Unauthenticated Denial of Service via Unbounded Topic Depth in Aedes MQTT Broker

CVE-2026-63341 is a high-severity Denial of Service (DoS) vulnerability in the Aedes MQTT broker for Node.js affecting all versions prior to 1.1.1. The flaw stems from missing input validation on the hierarchical depth of incoming MQTT topic strings before they are passed to the underlying topic-matching engine, qlobber. Because qlobber enforces a hardcoded limit of 100 topic levels and throws a synchronous exception when this threshold is exceeded, an attacker can send a single MQTT packet containing 101 or more forward-slash-separated levels to trigger an uncaught exception that immediately terminates the Node.js broker process.

Alon Barad
Alon Barad
3 views•12 min read
•about 4 hours ago•CVE-2026-61725
6.1

CVE-2026-61725: Path Traversal and Arbitrary File Overwrite in sevenz-rust2

CVE-2026-61725 is a path traversal vulnerability (CWE-22) in the sevenz-rust2 Cargo package prior to version 0.21.1. The vulnerability occurs because the decompression utility fails to validate or sanitize archive entry filenames before joining them with the destination directory path via std::path::Path::join. An attacker who supplies a crafted 7z archive containing parent directory traversal sequences or absolute paths can write or overwrite arbitrary files on the host filesystem with the privileges of the extracting process.

Alon Barad
Alon Barad
3 views•12 min read
•about 5 hours ago•CVE-2026-77309
8.8

CVE-2026-77309: Unauthenticated Remote Code Execution in NASA-AMMOS AIT-GUI via Incomplete File Cleanup and Path Traversal

CVE-2026-77309 (GHSA-jrjx-9jjw-747m) is a high-severity unauthenticated remote code execution vulnerability affecting NASA-AMMOS AIT-GUI versions 1.0.0 through 2.5.0. The flaw results from chaining an incomplete file cleanup error in the telemetry query endpoint (/tlm/query) with a directory traversal flaw in the script execution endpoint (/script/run). An attacker on an adjacent network can stage arbitrary Python statements into a static file on disk and execute that file under the privileges of the AIT-GUI web server process.

Amit Schendel
Amit Schendel
4 views•10 min read
•about 6 hours ago•CVE-2026-77332
2.3

CVE-2026-77332: Surfaced-Memory Envelope Injection and Stored Prompt Injection in memtomem-stm

A vulnerability classified under CWE-116 (Improper Encoding or Escaping of Output) exists in memtomem-stm versions 0.1.0 through 0.1.33. The software acts as a Model Context Protocol (MCP) short-term memory proxy gateway for AI agent architectures. When surfacing retrieved memories and scratchpad items into downstream Large Language Model (LLM) context envelopes, the SurfacingFormatter component fails to properly sanitize untrusted memory attributes. An attacker with the ability to plant untrusted content into the agent's long-term or working memory store can inject arbitrary structural delimiters, close the <surfaced-memories> XML boundary prematurely, visually obscure instructions using Unicode bidirectional overrides, and execute stored prompt injection attacks against downstream LLM agents.

Alon Barad
Alon Barad
7 views•10 min read
•about 7 hours ago•CVE-2026-77331
2.0

CVE-2026-77331: Race Condition and Session State Corruption in memtomem-stm

CVE-2026-77331 (GHSA-72jh-722p-rqr3) is a concurrency and session-state synchronization vulnerability in memtomem-stm, a short-term memory proxy gateway for AI agents. In versions 0.1.2 prior to 0.1.34, concurrent Model Context Protocol (MCP) Remote Procedure Call (RPC) failures or mid-flight task cancellations can initiate duplicate reconnect loops and permit dirty or cancelled transport sessions to be reused. This deficiency results in session-state integrity loss and local service degradation.

Amit Schendel
Amit Schendel
5 views•9 min read