Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessGain on-path network position (rogue AP/ARP spoof)
-
DeliveryVictim starts mcp-remote device-code OAuth flow
-
ExploitClient secret sent to http endpoint
-
ExecutionCapture secret and access/refresh tokens in cleartext
-
ImpactReplay tokens to impersonate client and access protected resources
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A04:2025 - Cryptographic Failures
ASVS 5.0
V12 - Secure Communication Transport encryption
|
| Exploitation Conditions | Exploitation requires THREE concrete conditions to coincide: (1) the target MCP/OAuth server's discovered device-authorization and token endpoints must be non-loopback http (cleartext) URLs - mcp-remote fails to enforce HTTPS in authorizeWithDeviceCode; (2) the attacker must occupy an on-path/man-in-the-middle network position to observe that traffic; and (3) the victim must actively initiate the device-code OAuth grant (passive user interaction) so that the client secret is transmitted. Limiting factors: loopback (localhost) endpoints and any server correctly using https are not vulnerable, and the attacker gains nothing without both the MITM vantage point and a live auth flow - this is not exploitable against a server configured with HTTPS endpoints. |
| Impact | This is a genuine but moderate-priority confidentiality flaw, correctly rated Medium (CVSS 4.0 base 6.0). The provided vector (AV:N/AC:L/AT:P/PR:N/UI:P/VC:H/VI:N/VA:N) tells a consistent story: network-reachable and unauthenticated (PR:N), but gated by two real-world dampeners - AT:P (a specific attack requirement, i.e. the attacker must hold an on-path/MITM position AND the OAuth server must advertise non-loopback http endpoints) and UI:P (passive user interaction, the victim must initiate the device-code auth flow). High confidentiality impact (VC:H) is justified because client secrets plus access and refresh tokens are exposed, but there is no direct integrity or availability impact (VI:N/VA:N). No EPSS score, KEV listing, or confirmed POC was provided: exploitation requires both a privileged network vantage point and a misconfigured/plaintext authorization server, which argues against broad opportunistic exploitation. A public technical write-up (hackmd) and a VulnCheck advisory exist, raising the chance of a working PoC, but no exploit code is confirmed in the input. Net: a real defect worth patching, especially for anyone running mcp-remote against OAuth servers reachable over untrusted networks, but not an emergency relative to CVSS-high network-RCE issues. |
| Exploit ScenarioDerived | A victim runs mcp-remote (0.8.0-0.14.3) to connect to a remote MCP server whose discovered device-authorization or token endpoint is a non-loopback http (cleartext) URL. An attacker positioned on the network path - e.g. on shared Wi-Fi or via ARP/DNS spoofing - passively captures the device-code exchange as the user authenticates, harvesting the client secret and the issued access and refresh tokens. No public weaponized exploit is confirmed, though a technical write-up detailing the flaw is publicly available. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – Oct 11, 2026
Cleartext transmission in mcp-remote 0.8.0 through 0.14.3 allows unauthenticated on-path attackers to capture OAuth client secrets and issued access/refresh tokens when the device-code authorization flow uses non-loopback HTTP endpoints. Exploitation requires a man-in-the-middle network position and a victim actively initiating the device-code grant; no public exploit code exists, and the issue is rated medium severity (CVSS 4.0 base 6.0).
RemediationAI
No vendor-released patch identified at time of analysis. As a workaround, ensure the OAuth server's device-authorization and token endpoints use HTTPS (or loopback addresses), and avoid using mcp-remote over untrusted networks when HTTP endpoints are configured. Monitor the GitHub repository (https://github.com/punkpeye/mcp-remote) and the VulnCheck advisory for a patched release. Trade-off: enforcing HTTPS may require server configuration changes but prevents credential theft; network-level protections such as VPNs or TLS inspection can partially mitigate but are not substitutes for proper endpoint security.
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
mcp-remote is a client library for connecting to Model Context Protocol (MCP) servers, implementing the OAuth device-code grant in authorizeWithDeviceCode. The function fails to enforce HTTPS for discovered device-authorization and token endpoints, allowing cleartext transmission of client credentials and tokens (CWE-319). The CPE cpe:2.3:a:punkpeye:mcp-remote:*:*:*:*:*:*:*:* indicates the affected package. The root cause is missing transport security validation, enabling network eavesdropping when endpoints are http URLs; source references include device-authorization.ts and client-credentials.ts.
Know when Mcp Remote 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:L/AT:P/PR:N/UI:P/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N
AC:H because the attacker needs an on-path position and an http-configured server; UI:R as the victim must start the device-code flow; C:H for secret/token disclosure with no integrity or availability impact.
Primary rating from Vendor (VulnCheck).
CVSS VectorVendor: VulnCheck
DescriptionCVE.org
mcp-remote 0.8.0 through 0.14.3 contains a cleartext transmission vulnerability in authorizeWithDeviceCode that sends client secrets and receives tokens without enforcing HTTPS endpoints. When discovered device authorization and token endpoints are non-loopback http URLs, on-path network attackers can capture the client secret plus issued access and refresh tokens.
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
Information disclosure occurs when an application unintentionally exposes sensitive data that aids attackers in reconnaissance or directly compromises security.
EUVD-2026-96990
GHSA-jrvg-vw6c-rp59
NVD
CVE.org
EPSS
Exploit-DB