GHSA-8vf2-hxg5-crqh: Kubeshark: MCP `start_kubeshark` argument injection via `pod_regex` → image redirect → cluster takeover
- Identifiers
- CVE-2026-129236GHSA-8vf2-hxg5-crqh
- Published
- Record updated
- Affected
- github.com/kubeshark/kubeshark >= 0.0.2-0.20260206182249-0bbbb473ea3f, < 0.0.2-0.20260812200004-a76a1920dc75
- Fixed in
- 0.0.2-0.20260812200004-a76a1920dc75
- released August 12, 2026, 96 days before disclosure
- Known exploitation
- Not listed in the CISA Known Exploited Vulnerabilities catalog at the last check.
Summary
The `start_kubeshark` MCP tool in `github.com/kubeshark/kubeshark` (versions `>= v53.1.0`, `< v53.4.0`) appends the caller-supplied `pod_regex` as a bare positional argument to a `kubeshark tap` command with no `--` separator, so a value beginning with `--set=` is parsed as a Helm flag. An attacker who can influence tool arguments, such as through indirect prompt injection against an agent driving `kubeshark mcp --allow-destructive`, can override `tap.docker.registry` and `tap.docker.tag` to deploy attacker-controlled images into a privileged DaemonSet with `hostNetwork` and the host filesystem mounted, leading to cluster-admin takeover. Kubeshark rates it 7.2 High, reducing the reporter's 8.3 on attack vector and privileges-required grounds.
Mitigation
Fixed in v53.4.0.
Topics
Related items
- MediumCVE-2026-126681: AWS Labs Timestream for InfluxDB MCP Server exposes API tokenSimilar attack · NVD/CVE Database
- HighCVE-2026-126681 - Improper restriction of intended endpoints in AWS Labs Timestream for InfluxDB MCP ServerSimilar attack · AWS Security Bulletins
- Medium'Sleeper Agent' Tactic Hides Payloads in Open-Weight AI ModelsSimilar attack · Dark Reading
- MediumCursor Supply Chain Attack Started at Fireworks AISimilar attack · SecurityWeek
- MediumAmazon Uncovers AI Agent Campaign Targeting LiteLLM, Dify, and Other Apps for API KeysSimilar attack · The Hacker News