Three Major Security Flaws in Anthropic's Git Server: Implications for North East India

Vulnerabilities Discovered in Anthropic's mcp-server-git

Recent findings by Cyata Researcher Yarden Porat have uncovered three significant security vulnerabilities in Anthropic's mcp-server-git, a Python package and an Artificial Intelligence (AI) tool used for reading, searching, and manipulating Git repositories programmatically via large language models (LLMs). These vulnerabilities, if exploited, could potentially allow attackers to read or delete arbitrary files and execute code on a system.

Path Traversal and Argument Injection Vulnerabilities

The three identified vulnerabilities are CVE-2025-68143, CVE-2025-68144, and CVE-2025-68145. The first, CVE-2025-68143, is a path traversal vulnerability that arises due to the git_init tool accepting arbitrary file system paths during repository creation without validation. The second, CVE-2025-68144, is an argument injection vulnerability that occurs due to git_diff and git_checkout functions passing user-controlled arguments directly to Git CLI commands without sanitization. The third, CVE-2025-68145, is another path traversal vulnerability that arises due to a missing path validation when using the --repository flag.

Remote Code Execution Scenarios

Successful exploitation of these vulnerabilities could potentially allow an attacker to turn any directory into a Git repository, overwrite any file with an empty diff, and access any repository on the server. In a documented attack scenario, the three vulnerabilities could be chained to write to a ".git/config" file and achieve remote code execution by triggering a call to git_init.

Relevance to North East India and Broader Indian Context

Given the increasing reliance on AI and digital tools in various sectors across India, including North East India, these vulnerabilities pose a potential threat. Organizations and individuals using mcp-server-git should ensure they have updated to the latest version (2025.9.25 or 2025.12.18) to mitigate these risks.

Reflections and Future Implications

These vulnerabilities serve as a reminder of the importance of robust security measures, especially in reference implementations like Anthropic's mcp-server-git. The entire MCP ecosystem may need deeper scrutiny to ensure the security of AI-based tools used in various sectors.