CVE-2026-73498: MCP Atlassian arbitrary file read in attachment upload

CVE-2026-73498 · Severity: high · CVSS 7.7 · Published 2026-08-12

MCP Atlassian is a server that allows AI agents and clients to integrate with Atlassian products like Confluence and Jira. A flaw in the file upload feature allows authenticated users to read any file accessible to the server process and steal it by uploading it as an attachment, potentially exposing sensitive credentials like API tokens stored in environment variables.

Technologies: mcp-atlassian (PyPI). Vendors: PyPI.

Quick answers

Is CVE-2026-73498 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-73498 being exploited in the wild.
How severe is CVE-2026-73498?
High severity, with a CVSS score of 7.7, published 12 August 2026.
Which products and versions are affected?
mcp-atlassian (PyPI) ECOSYSTEM:introduced=0,fixed=0.22.0.
Is there a fix for CVE-2026-73498?
Yes. mcp-atlassian (PyPI): upgrade to 0.22.0 or later. Junglewise finds every place you run it, drafts the upgrade and applies it once you approve.
How is CVE-2026-73498 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-73498?
Check whether you run mcp-atlassian (PyPI) at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

How to fix

  • mcp-atlassian (PyPI): upgrade to 0.22.0 or later
    pip install "mcp-atlassian>=0.22.0"

Technical details

The vulnerability is a path traversal / arbitrary file read in the confluence_upload_attachment function. The vulnerable code in src/mcp_atlassian/confluence/attachments.py passes client-supplied file_path directly to open(file_path, "rb") via _upload_attachment_direct() without calling validate_safe_path, allowing absolute paths or ../ traversal sequences. An authenticated MCP client can supply any file path accessible to the server process, causing the server to read and exfiltrate that file to Confluence as an attachment. If an AI agent can be induced to invoke this tool through untrusted content, attacker-controlled prompts can disclose sensitive environment variables such as CONFLUENCE_API_TOKEN. The fix in version 0.22.0 adds validate_safe_path validation to confine uploads to the working directory before file access.

Affected products

  • sooperset mcp-atlassian prior to 0.22.0

Identifiers

  • CVE-2026-73498
  • GHSA-g5r6-gv6m-f5jv
  • PYSEC-2026-3683

Timeline

  • 2026-08-12: disclosed
  • 2026-07-10: patched: Fix merged in version 0.22.0

References

This page as Markdown. Free to reuse under CC BY 4.0 with a link back.