CVE-2026-55422: conda-forge.org: Stored DOM XSS via unsanitized dangerouslySetInnerHTML in the status dashboard
conda-forge is a GitHub organization containing repositories of conda recipes. In commits 85748ad2368a and prior, two React components render externally-sourced strings directly as HTML using dangerouslySetInnerHTML, with no sanitization (the project has no DOMPurify or equivalent dependency). JSON files are produced automatically by the conda-forge autotick bot (regro/cf-scripts) while it processes every conda-forge feedstock and its upstream source artifacts. The error/status strings embed raw output captured from those processes (build logs, patch output, tracebacks). The website renders them verbatim as HTML. At time of publication, there are no publicly available patches.
Affected Software
Event History
Frequently Asked Questions
Who can realistically exploit this issue?
An attacker needs to cause attacker-controlled content to appear in data processed by the conda-forge autotick bot, such as upstream source artifacts or output that is captured in build logs, patch output, or tracebacks. Exploitation also requires a user to view a status-dashboard page that renders the affected string.
Is authentication required to trigger the vulnerable rendering?
The vulnerability has low-privileges-required and user-interaction-required characteristics. The affected site renders automatically produced JSON, but the available information does not establish that an attacker needs direct access to conda-forge.org.
Are patches available?
No publicly available patches existed at the time of publication. The affected code includes commits 85748ad2368a and earlier.
What can be done while no patch is available?
Avoid rendering externally sourced error or status strings as HTML; treat them as text instead. Where HTML rendering is necessary, sanitize the strings before passing them to dangerouslySetInnerHTML.