Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessObtain authenticated Flowise session (PR:L)
-
DeliveryCraft Custom MCP node with malicious npx package in mcpServerConfig
-
Exploitnpx fetches attacker-controlled npm package
-
ExecutionPackage code runs as Flowise server process
-
ImpactFull host compromise / data exfiltration
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A05:2025 - Injection
ASVS 5.0
V1 - Encoding and Sanitization OS command sanitization, avoid shell exec
MITRE ATT&CK
2 techniques across 1 tactic: Execution The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Requires an authenticated Flowise account (PR:L) with the ability to create or edit a Custom MCP node and control the mcpServerConfig parameter; the deployment must have the MCP/npx execution path reachable (npx/npm present on the host, which is standard for Flowise's Node.js runtime). The exploit works against Flowise versions before 3.1.4. Limiting factors: authentication is mandatory (no pre-auth exploitation), so exposure is driven by how many users can reach the node editor and by whether the instance is internet-facing or has weak/shared credentials; there is no user-interaction requirement beyond the attacker's own actions. |
| Impact | This is a genuine, high-priority issue but with a meaningful gating condition. The vendor CVSS 4.0 vector (AV:N/AC:H/AT:N/PR:L/UI:N with full VC:H/VI:H/VA:H) describes authenticated (PR:L) remote code execution - the PR:L is corroborated by the description's explicit 'authenticated attackers' wording. Exploitation status is 'no public exploit identified at time of analysis' - there is no CISA KEV listing and no POC signal in the provided data, so this should not be treated as actively exploited despite the 9.0 score. EPSS is not provided, so probabilistic exploitation likelihood cannot be quantified. The reporting by VulnCheck plus a published GHSA advisory and available vendor patch indicate a well-substantiated, real vulnerability rather than a speculative or over-scored one. The one contestable metric is AC:H: once an attacker holds a valid authenticated session, supplying an attacker-controlled npx package name in mcpServerConfig is mechanically simple, so the attack complexity is arguably low rather than high - this would push the true severity even higher, not lower. The subsequent-system impacts (SC:H/SI:H/SA:H) in the vendor 4.0 vector appear overstated: code executes within the Flowise server process context with no clear scope change to a separate security authority, so a same-system (SC:N/SI:N/SA:N) reading is more defensible. Net: a real priority to patch for any internet-exposed or multi-user Flowise instance, tempered by the authentication requirement. |
| Exploit ScenarioDerived | An attacker with valid low-privilege credentials to a Flowise instance (or one obtained via weak/default auth or credential theft) creates or edits a Custom MCP node and sets mcpServerConfig to invoke npx against an attacker-published npm package. When Flowise launches the MCP server, npx downloads and executes that package's code on the server, yielding arbitrary command execution in the Flowise process context. No public POC is identified in the provided data, but the technique (malicious npm package + npx) is well-understood and trivially reproducible once authenticated. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Sep 16, 2026
Authenticated remote code execution in Flowise versions before 3.1.4 lets a low-privileged user who can create or edit a Custom MCP node run arbitrary code on the Flowise server by supplying attacker-controlled npm package names in the mcpServerConfig parameter, which Flowise then invokes through npx. The issue carries a vendor-published CVSS 4.0 score of 9.0, while our independent assessment rates it CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H (8.8), reflecting that the attacker must first hold a valid Flowise account with access to the node editor - no pre-authentication path was identified. No public exploit code was identified at time of analysis and the vulnerability has not been confirmed actively exploited (no CISA KEV listing), so risk is driven mainly by how many users can reach the MCP node configuration UI and whether the instance is internet-facing or uses weak or shared credentials.
RemediationAI
Upgrade Flowise to 3.1.4 or later, which is the vendor-released fix and the only complete remediation; confirm the running build after upgrade and reference the vendor advisory at https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-vcwp-f9rq-3887 and the VulnCheck advisory at https://www.vulncheck.com/advisories/flowise-before-3.1.4-remote-code-execution-via-custom-mcp-npx. Where immediate upgrade is not possible, apply layered compensating controls with awareness of their trade-offs: restrict Custom MCP node creation and editing to a tightly trusted administrator group and revoke node-editor privileges from ordinary accounts, since the flaw is only reachable by an authenticated user with that capability (removes the attack path but limits legitimate MCP authoring); remove or disable the Custom MCP node from the node palette if MCP functionality is unused (breaks all MCP-based flows); remove npx/npm from the PATH of the Flowise process or run Flowise in a container image without the npm CLI (mitigates this execution path but breaks any MCP server that Flowise is expected to launch via npx); and enforce outbound egress allowlisting or a private npm proxy so the process cannot fetch arbitrary public packages (attacker can no longer pull their payload, though a package already registry-cached or reachable through the proxy may still execute). Supplement with runtime process monitoring and alerting on Flowise spawning npx or unexpected child processes, and audit existing Custom MCP node configurations for suspicious or unrecognized package names, treating any such entries as potential compromise requiring incident response. Note there is no vendor-released patch status ambiguity here: the fix version is 3.1.4, and no public exploit code was identified at time of analysis.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 1 package in euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
The flaw is an OS command injection (CWE-78) in Flowise's Model Context Protocol integration, specifically the Custom MCP node. Flowise is a Node.js/TypeScript low-code LLM orchestration platform, and its MCP nodes are designed to launch MCP servers locally, typically by shelling out to npx with a package specifier taken from the mcpServerConfig parameter. Because the package name is not validated against an allowlist or sanitized before being passed to the process-spawning layer, an authenticated user with node-editor privileges can substitute an arbitrary npm package (including one they publish themselves) and have Flowise download and execute it on the host under the Flowise process account. This is a configuration-to-execution path rather than a memory-safety bug: the injection surface is a design assumption that the operator supplies trustworthy tooling, and the standard Flowise deployment already has npx/npm on the host and outbound registry access as normal runtime prerequisites. CPE records (cpe:2.3:a:flowiseai:flowise) confirm the affected product identity, and ENISA's EUVD entry (EUVD-2026-78789) tracks the affected range as Flowise 0 through 3.1.3.
Know when Flowise gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H
Authenticated (PR:L) network RCE that is mechanically simple once logged in (AC:L); code runs in-process with no scope change (S:U), yielding full C/I/A impact.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
Lifecycle Timeline
5DescriptionCVE.org
Flowise before 3.1.4 contains a remote code execution vulnerability in the Custom MCP node that allows authenticated attackers to execute arbitrary code by supplying npx package names in the mcpServerConfig parameter. Attackers can invoke npx with attacker-controlled npm packages to execute code on the Flowise server.
Articles & Coverage 2
Technical POC Analysis
The flaw is OS command execution via an unsafe executable allowlist (CWE-78; the advisory also lists CWE-94).
Full technical POC analysis - root cause, step-by-step exploitation and patch breakdown - is part of the Pro and Business plans.
Executive ViewAI
Flowise, an AI workflow orchestration platform, contains a critical authenticated remote code execution vulnerability (CVE-2026-91931) affecting versions prior to 3.1.4. A low-privileged user with access to the Custom MCP node editor can execute arbitrary code on the server, potentially leading to full system compromise, data theft, or lateral movement. Organizations using Flowise should prioritize patching, especially if the instance is internet-facing or accessible to many users, as the risk increases with the number of accounts that can reach the vulnerable interface.
Apply vendor patch.
Threat Intelligence Overview
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
OS command injection occurs when an application passes unsanitized user input directly into system shell commands.
Remote Code Execution represents the critical moment when an attacker successfully runs arbitrary code on a target system without physical access.
Flowise version 3.0.5 contains a remote code execution vulnerability in the CustomMCP node. The mcpServerConfig paramete
Flowise versions before 3.0.1 allow unauthenticated access to the Custom MCPs feature, which is designed to execute OS c
FlowiseAI Flowise version 2.2.6 contains an arbitrary file upload vulnerability in the /api/v1/attachments endpoint. Una
Arbitrary file read and write in Flowise versions prior to 3.0.8 allows an authenticated attacker with access to the flo
Remote code execution in Flowise before 3.1.2 allows any authenticated user (or API caller with chatflow view/update per
Flowise is a drag & drop user interface to build a customized large language model flow. Rated critical severity (CVSS 9
Authenticated users can upload arbitrary files to Flowise 3.0.7 due to missing validation, enabling persistent storage o
Unrestricted file upload in Flowise LLM workflow builder before 3.0.13 via /api/v1/attachments endpoint allows unauthent
Missing authentication on NVD data endpoint in Flowise before 3.0.13 allows unauthenticated access to internal vulnerabi
Authenticated remote code execution in FlowiseAI Flowise (v3.0.1 up to but not including 3.0.8, and later versions when
Remote code execution in Flowise before 3.1.3 arises because its Pyodide Python-code validator (validatePythonCodeForDat
Remote code execution in FlowiseAI Flowise before 3.1.3 lets an attacker execute arbitrary OS commands as the Flowise pr
EUVD-2026-78789
GHSA-q5vh-j6x8-9f53
NVD
CVE.org
EPSS
Exploit-DB