Security Advisory - Published 2026-07-11 - DevOps / AI Tooling
DevOps and AI tooling batch: check exposed dashboards, Docker access, Kubernetes tools, and outbound logs
This batch is about internal tools that often sit close to credentials: local Kubernetes dashboards, document conversion workers, MCP tools, crawler jobs, AI detector services, and monitoring APIs.
Affected CVEs in this batch
| CVE | Product | Affected | Review | CVSS |
|---|---|---|---|---|
| CVE-2026-61459 | MCP Server Kubernetes | < 3.9.0 | kubectl tool calls, service account scope, audit logs, and cluster namespaces | 9.8 |
| CVE-2026-15378 | guardrails-detectors | vendor advisory | detector endpoints, outbound requests, proxy logs, and tenant boundaries | 9.3 |
| CVE-2026-56261 | Crawl4AI | < 0.8.7 | Docker API exposure, job webhooks, outbound requests, and worker logs | 9.2 |
| CVE-2026-55884 | Tilt | 0.20.8 through 0.37.3 | HUD network binding, handler exposure, workspace logs, and cluster permissions | 9.2 |
| CVE-2026-55882 | Tilt | 0.19.5 through 0.37.3 | HUD exposure, profiling endpoints, developer workstations, and cluster context | 8.3 |
| CVE-2026-55883 | Tilt | 0.24.0 through 0.37.3 | HUD WebSocket access, CSRF behavior, local networks, and cluster actions | 8.3 |
| CVE-2026-55229 | Gotenberg | < 8.34.0 | conversion jobs, LibreOffice logs, temporary files, and container egress | 7.5 |
| CVE-2026-59155 | Nezha Monitoring | < 2.2.5 | DDNS settings, notification endpoints, API logs, and monitor credentials | 6.9 |
What to check
- Tilt HUD binding, local network exposure, workspace logs, and Kubernetes context permissions.
- Gotenberg document conversion queues, LibreOffice logs, temporary files, and container egress.
- MCP Server Kubernetes structured tool calls, service account permissions, namespace scope, and audit logs.
- Crawl4AI Docker API access, job webhooks, worker logs, and outbound request logs.
- guardrails-detectors endpoints, proxy logs, tenant boundaries, and internal network reachability.
- Nezha Monitoring API logs, DDNS settings, notification configuration, and monitor credentials.
Safe fix path
- Remove public exposure from developer dashboards, MCP transports, Docker APIs, and monitoring endpoints.
- Patch affected tools before reusing old workspaces or worker containers.
- Preserve tool logs, Kubernetes audit logs, container logs, proxy logs, and job histories.
- Rotate tokens, kubeconfigs, webhook secrets, and monitor credentials if tool integrity is uncertain.
Compromise indicators
- Unexpected Kubernetes reads or writes, new namespaces, changed workloads, or service-account token use.
- Document conversion jobs that reference unknown documents, URLs, or output paths.
- Crawler or AI detector traffic to internal addresses that does not match normal jobs.
- Monitoring API reads, notification changes, or DDNS changes outside a real maintenance window.
When to ask Ping7 for repair
Use Ping7 CVE Repair when a developer tool had broad credentials, Kubernetes access, Docker access, or internal network reachability and you need logs reviewed before restarting jobs.