Description

A security flaw has been discovered in cosmicstack-labs mercury-agent up to 1.1.13. The impacted element is the function PermissionManager.checkShellCommand of the file mercury-agent/src/capabilities/permissions.ts of the component Shell Command Permission Check. Performing a manipulation results in improper authorization. The attack can be initiated remotely. The exploit has been released to the public and may be used for attacks. The project was informed of the problem early through an issue report but has not responded yet.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "6.3",
  "pubDate": "2026-09-14T19:18:11.480Z",
  "pubdate": "2026-09-14T19:18:11.480Z",
  "executiveSummary": "A critical improper authorization vulnerability exists in the mercury-agent component of cosmicstack-labs, specifically within the PermissionManager.checkShellCommand function.\nThe vulnerability allows remote attackers to execute unauthorized shell commands, effectively bypassing security controls meant to restrict system access.\nAffected versions include mercury-agent up to and including 1.1.13.\nThis flaw presents a high risk as it enables unauthorized system interaction, potentially leading to full compromise of the host environment depending on the execution context.\nThe vulnerability is currently exploitable, with proof-of-concept exploits available publicly.\nDespite early disclosure via the issue reporting process, no patch or official response from the maintainers has been provided, leaving deployments exposed to remote exploitation.",
  "technicalDetails": "The vulnerability resides in mercury-agent/src/capabilities/permissions.ts, specifically within the logic implemented in PermissionManager.checkShellCommand.\nThe root cause is a failure to sufficiently validate or constrain shell command input, leading to an improper authorization state where the component fails to enforce configured security policies before passing commands to the underlying operating system shell.\nBecause the function is reachable via remote interfaces, an attacker can supply malicious or unauthorized command strings that bypass the intended restriction mechanism.\nThe attack flow commences when a remote attacker interfaces with the mercury-agent service, providing a payload designed to trigger the Shell Command Permission Check component.\nUpon receiving the request, the PermissionManager.checkShellCommand function performs a faulty validation check. Due to the flaw, the validation logic incorrectly authorizes the execution of arbitrary commands, even if those commands are not explicitly permitted by the agent configuration.\nOnce authorized, the command is passed to the system's execution environment, where it is executed with the privileges of the mercury-agent process.\nThis allows for remote command execution (RCE) on the host machine. Post-exploitation impact includes unauthorized system enumeration, data exfiltration, or the establishment of persistence mechanisms, contingent upon the service's operating privileges.\nAs the exploit code is publicly available, the barrier to entry for attackers is minimal, requiring only the ability to reach the service over the network.\nThe lack of input sanitization or robust permission verification within the checkShellCommand function represents a significant security design failure that allows for unintended interaction with the system shell."
}