CVE-2025-66401: MCP Watch has a Critical Command Injection in cloneRepo allows Remote Code Execution (RCE) via malicious URL
Summary The MCPScanner class contains a critical Command Injection vulnerability in the cloneRepo method. The application passes the user-supplied githubUrl argument directly to a system shell via execSync without sanitization. This allows an attacker to execute arbitrary commands on the host machine by appending shell metacharacters to the URL.
Published Dec 2, 2025
Affected packages
mcp-watch
- Ecosystem
- npm
- Affected
- <= 0.1.2
- Fixed in
- No patched version yet
- Check
npm ls mcp-watch
Advisory details
From the GitHub Advisory Database (CC BY 4.0).
Summary
The MCPScanner class contains a critical Command Injection vulnerability in the cloneRepo method. The application passes the user-supplied githubUrl argument directly to a system shell via execSync without sanitization. This allows an attacker to execute arbitrary commands on the host machine by appending shell metacharacters to the URL.
Details
The vulnerability exists in the src/scanner/MCPScanner.ts file within the cloneRepo method.
The code uses child_process.execSync to execute a git clone command:
Because execSync spawns a shell (defaulting to /bin/sh on Unix or cmd.exe on Windows), any shell metacharacters present in the url argument will be interpreted by the shell. The application does not validate that the url is a valid Git URL, nor does it sanitize input for shell metacharacters.
PoC
Install the package or clone the repository.
Run the scanner using the CLI (or invoke scanRepository programmatically).
Provide a malicious URL containing a command separator (e.g., ;, &, or |) and a system command. payload : npm run scan:github "https://github.com/kapilduraphe/mcp-watch & calc.exe"
Impact
Severity: Critical
CVSS Score: 9.8 (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H) Description: This vulnerability allows an attacker to execute arbitrary code on the machine running the scanner.
If run by a developer locally, it compromises their workstation.
If deployed as a hosted scanning service, it grants the attacker full control over the server (RCE), leading to potential data exfiltration, service disruption, or further lateral movement within the infrastructure.
Context Dependent Risk:
Local CLI : If you run this tool locally on your own machine, you are "hacking yourself." The risk is limited unless you copy-paste a malicious URL sent by someone else (e.g., "Hey, check this repo scan: npm run scan "https://git./..; rm -rf /").
Web Service / CI Pipeline (Critical Risk): If this scanner is deployed as a web service (e.g., "Paste your repo URL to scan"), an attacker can take full control of the server immediately.
Credited to viralvaghela.
References
How this page is made: the data panel comes from the GitHub Advisory Database, CISA’s KEV catalog, the NVD and FIRST EPSS, and updates twice a day. No analysis has been written for this entry yet. Methodology
More Other MCP servers and tools vulnerabilities
All Other MCP servers and tools entriesCVE-2026-77361
obsidian-tc : Symlink read-ACL scope bypass via lexical path checks
medium4.7Oct 23, 2026CVE-2026-114126
Storybook dev server vulnerable to code injection from the local network
high8.7Oct 22, 2026CVE-2026-113731
djust: Arbitrary absolute-path file read in the find_handlers_for_template MCP tool
low3.3Oct 21, 2026CVE-2026-114125
Storybook: Missing origin validation in Storybook preview allows code injection from malicious websites
critical9.4Oct 21, 2026CVE-2026-111770
MCP TypeScript server SDK: Shared transport or server instances delivered one HTTP client's messages to another
high7.7Oct 20, 2026GHSA-953h-535j-jg2v
agent-router: MCP session resume does not verify the embedded anti-hijacking subject; default Helm seed makes session IDs forgeable
medium6.8Oct 19, 2026
