Skip to main content

CVE-2024-6483: Aimstack Aim Path Traversal Vulnerability

CVE-2024-6483 is a path traversal vulnerability in Aimstack Aim that enables attackers to delete arbitrary files or directories through the runs/delete-batch endpoint. This article covers technical details, affected versions, impact assessment, and mitigation strategies.

Published:

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

CVE-2024-6483 Overview

CVE-2024-6483 is a path traversal vulnerability in aimhubio/aim version 3.19.3, an open-source experiment tracking tool used in machine learning workflows. The flaw resides in the runs/delete-batch API endpoint, which accepts user-supplied run names to identify log and metadata files for deletion. The endpoint fails to sanitize or canonicalize these inputs, allowing an attacker to supply traversal sequences such as ../ to reach files outside the intended directory. Successful exploitation deletes arbitrary files or directories on the host running the aim server, resulting in data loss or denial of service.

Critical Impact

Unauthenticated network attackers can delete arbitrary files or directories accessible to the aim process, disrupting experiments and corrupting ML tracking data.

Affected Products

  • aimstack aim 3.19.3 (Python package)
  • Deployments exposing the runs/delete-batch endpoint over the network
  • Downstream environments bundling the vulnerable aim server component

Discovery Timeline

  • 2025-03-20 - CVE-2024-6483 published to the National Vulnerability Database (NVD)
  • 2026-06-17 - Last updated in the NVD database

Technical Details for CVE-2024-6483

Vulnerability Analysis

The vulnerability is classified as Relative Path Traversal under [CWE-23]. The runs/delete-batch endpoint receives a list of run identifiers from the client and uses those identifiers to construct filesystem paths for the associated log and metadata files. The server concatenates the untrusted input into a path and invokes deletion routines without verifying that the resolved path remains within the designated storage directory.

An attacker sends a crafted request containing run names such as ../../etc/important.conf. The server resolves the string relative to its data directory and deletes the target. Because the endpoint does not require authentication by default in exposed deployments, any network-reachable attacker can issue the request. The impact is limited to file integrity and availability, not confidentiality, since the operation only deletes rather than reads data.

Root Cause

The root cause is missing input validation on the run-names parameter. The handler trusts the client to supply well-formed run identifiers and does not canonicalize the resulting path or compare it against an allowed base directory. Standard mitigations such as os.path.realpath checks or allowlist validation against known run IDs are absent.

Attack Vector

Exploitation requires network access to the aim web server and knowledge of the runs/delete-batch endpoint. The attacker submits a POST request containing traversal sequences in the batch of run names. No privileges or user interaction are required. Attackers can chain repeated requests to remove configuration files, system binaries, or other experiment data accessible to the service account running aim.

No verified public exploit code is available. See the Huntr Bounty Listing for additional technical details.

Detection Methods for CVE-2024-6483

Indicators of Compromise

  • HTTP requests to /api/runs/delete-batch containing ../, ..\, URL-encoded %2e%2e%2f, or absolute path prefixes in the run-name payload
  • Unexpected deletion of files or directories outside the aim data directory, visible in filesystem audit logs
  • aim server process errors referencing missing files that were not previously tracked by the application

Detection Strategies

  • Inspect reverse-proxy and application logs for POST requests to runs/delete-batch with suspicious characters in the request body
  • Enable filesystem auditing (auditd on Linux) on directories hosting the aim service to flag deletions outside the expected run storage path
  • Compare the aim data directory inventory against expected run IDs and alert on orphaned deletions

Monitoring Recommendations

  • Baseline normal API traffic patterns to the aim server and alert on anomalous deletion volume or payload sizes
  • Forward aim application logs and host filesystem events to a centralized analytics platform for correlation
  • Review service-account file access scope and alert when the aim process touches files outside its designated directory

How to Mitigate CVE-2024-6483

Immediate Actions Required

  • Restrict network access to the aim server using firewall rules or an authenticated reverse proxy until a patched release is deployed
  • Run the aim server under a dedicated, unprivileged service account with write access limited to its data directory
  • Disable or block the runs/delete-batch endpoint at the proxy layer if deletion functionality is not required

Patch Information

No patched version is documented in the available vulnerability references. Monitor the Huntr Bounty Listing and the upstream aimhubio/aim repository for a fixed release addressing CVE-2024-6483. Upgrade immediately once a corrected version is published.

Workarounds

  • Place aim behind an authenticating reverse proxy and require a valid session for all API endpoints
  • Add a web application firewall (WAF) rule to reject requests to runs/delete-batch containing .., /, \, or URL-encoded equivalents in run-name fields
  • Mount the aim data directory on a dedicated volume and apply mandatory access controls (SELinux, AppArmor) to prevent the process from deleting files outside that volume
bash
# Example nginx rule to block traversal payloads to the vulnerable endpoint
location /api/runs/delete-batch {
    if ($request_body ~* "(\.\./|\.\.\\|%2e%2e%2f|%2e%2e/)") {
        return 403;
    }
    proxy_pass http://aim_backend;
}

Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

Sources: National Vulnerability Database, CISA KEV, Exploit-DB

Experience the Most Advanced Cybersecurity Platform

See how the world’s most intelligent, autonomous cybersecurity platform can protect your organization today and into the future.