Path Traversal in NousResearch Hermes-Agent

Compliance Impact

The vulnerability allows remote attackers to perform path traversal and access files outside the intended skills directory, potentially exposing sensitive files such as configuration files (.env) that may contain secrets like API keys and tokens.

Exposure of sensitive data due to this vulnerability could lead to non-compliance with data protection regulations such as GDPR and HIPAA, which require protection of personal and sensitive information from unauthorized access.

Therefore, if exploited, this vulnerability could compromise confidentiality requirements mandated by these standards, potentially resulting in regulatory violations and associated penalties.

Useful 0 Not Useful 0

Executive Summary

This vulnerability is a path traversal flaw in the Hermes-agent software, specifically in the skill_view function of the skills_tool.py file. An attacker can manipulate the name argument to access files outside the intended skills directory by using path traversal sequences like "../outside-skill" or absolute paths. This happens because the name parameter was joined with the trusted skills directory path without proper validation, allowing unauthorized access to files outside the trusted directory.

The vulnerability allows an attacker to bypass security checks and read arbitrary files on the system, including sensitive files such as configuration files (.env). The issue was fixed by adding a validation function that blocks path traversal attempts and rejects absolute paths, ensuring file access is restricted to the intended skills directory.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability can impact you by allowing a remote attacker to read sensitive files outside the intended directory of the Hermes-agent. By exploiting the path traversal flaw, attackers can access confidential information such as API keys, tokens, configuration files, and other secrets stored on the system.

Such unauthorized file disclosure can lead to further compromise of your system or application, as attackers may gain access to credentials or configuration details that facilitate additional attacks.

Useful 0 Not Useful 0

Detection Guidance

This vulnerability involves path traversal via manipulation of the `name` argument in the `skill_view` function of the Hermes agent. Detection can focus on monitoring for suspicious requests or inputs containing path traversal sequences such as "../" or absolute paths targeting the `skill_view` functionality.

You can detect attempts by searching logs or network traffic for requests to the Hermes agent that include path traversal patterns in the `name` parameter.

• Use grep or similar tools to search Hermes agent logs for path traversal patterns, e.g.: `grep -E "\.\./|/|C:\\" /path/to/hermes-agent/logs/*`
• Monitor network traffic for suspicious HTTP requests containing `../` sequences targeting the Hermes agent endpoints.
• If you have access to the Hermes agent source or runtime environment, you can audit calls to the `skill_view` function or instrument it to log suspicious `name` arguments.

Useful 0 Not Useful 0

Mitigation Strategies

The primary mitigation step is to apply the official patch that fixes the path traversal vulnerability in the Hermes agent. This patch introduces validation to reject path traversal sequences and absolute paths in the `name` argument.

Until the patch is applied, you should restrict access to the Hermes agent service to trusted users or networks to reduce the risk of exploitation.

Additionally, monitor logs and network traffic for exploitation attempts and consider implementing application-level filtering to block requests containing suspicious path traversal patterns.

• Apply the patch identified by commit 56f833efa427ccb444c0f9ad1759af1012f2124d as soon as possible.
• Restrict network access to the Hermes agent to trusted sources.
• Monitor logs and network traffic for path traversal attempts targeting the `skill_view` function.

Useful 0 Not Useful 0