CVE-2026-40179 is a stored cross-site scripting vulnerability in Prometheus 3.0 through 3.5.1 and 3.6.0 through 3.11.1. Multiple Prometheus web UI components insert attacker-controlled metric names and label values into innerHTML without HTML escaping. Affected rendering paths include Graph-page chart tooltips in both the Mantine and legacy React UIs, plus Metric Explorer fuzzy-search results and heatmap-cell tooltips in the legacy React UI. Prometheus 3.x permits UTF-8 metric and label names by default, allowing HTML-significant characters in these values. An attacker who can cause crafted metrics or labels to be stored can trigger script execution when a user views an affected UI element.
Mallory correlates every CVE against your assets, your vendors, and active adversary campaigns. Know which vulnerabilities matter for you, not just which ones are loud.
What it means. What to do now. Patch path, mitigations, and the assume-compromise checklist.
What an attacker gets, and what they’ve been doing with it.
If you can’t patch tonight, do this now.
Patch, then assume compromise.
1 valid exploit after Mallory filtered fakes, detection scripts, and README-only repos.
This repository is a small standalone proof-of-concept exploit for CVE-2026-40179, a stored XSS vulnerability in Prometheus 3.x. It contains only two files: a README with exploitation, cleanup, and testing guidance, and a single Python script that performs the exploit. The Python code manually encodes a Prometheus remote_write protobuf WriteRequest without generated protobuf bindings, compresses it with Snappy, and POSTs it with requests to the Prometheus remote_write endpoint at https://<target>:9090/api/v1/write. The injected time series uses the __name__ label to store an HTML/JavaScript payload in the metric name: an img tag with an onerror handler that executes JavaScript when the metric is rendered unsafely in the Prometheus web UI. The default payload is low-noise and only calls console.log, but the README explains how to swap it for alert(). The exploit capability is therefore persistent data injection leading to stored client-side code execution in the browser of a user viewing the affected metric. It is not a scanner or detector; it actively writes malicious series data to the target TSDB. The exploit requires a reachable vulnerable Prometheus instance, especially one exposing unauthenticated remote_write, and user interaction in a vulnerable UI surface to trigger execution. The README also documents cleanup endpoints under the optional admin API and notes that the artifact persists if admin cleanup is unavailable.
Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.
Vendor-confirmed product mapping. Mallory continuously reconciles this list against your asset inventory.
7 sources tracked across advisories and community write-ups. News coverage will land here when it surfaces.
No news coverage yet. Advisories and community discussion only.
Query your assets running an affected version, and investigate the blast radius.
Every observed campaign linking this CVE to a named adversary.
Malware families riding this exploit, with evidence and IOCs.
YARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Cross-references every affected SKU, including bundled OEM variants.
Community discussion across Reddit, Mastodon, and other social sources.