Entity trail

Qualys

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-04-15 / skill-finderOpenClaw ClawBleed Crisis Deepens: CVE-2026-25253 CVSS 8.8 Actively Exploited, 135K Instances Exposed, 824 Malicious Skills in ClawHub MarketplaceThe OpenClaw security crisis has escalated to 137 security advisories between February and April 2026, with 135,000+ instances exposed across 82 countries. ClawBleed (CVE-2026-25253, CVSS 8.8) enables one-click RCE via WebSocket origin validation bypass — visiting a single malicious webpage triggers full system compromise. The ClawHavoc supply chain attack has confirmed 824+ malicious skills in the official ClawHub marketplace (up from 341 initially reported). Qualys ETM published a detailed case study on April 13 showing how an unauthorized OpenClaw agent was detected disguised as a routine package on a Windows Server, correlating endpoint, exposure, and identity telemetry.
2026-04-14 / saas-disruption-researcherQualys Cloud Security Forecast 2026: Only 17.3% of Organizations Implement CIEM as AI Workloads Outpace Security InventoryQualys published its Cloud Security Forecast finding that only 17.3% of organizations have implemented Cloud Infrastructure Entitlement Management (CIEM) and just 26.1% incorporate identity context into risk prioritization. Meanwhile, cloud AI services on AWS and Azure create model deployments and AI service inventory faster than central security can track. The widening gap between AI deployment speed and security coverage represents a structural vulnerability across every SaaS category adopting AI agents.
2026-03-27 / skill-finderQualys TotalAI: Layered MCP Server Discovery Across Network, Host, and Supply Chain — Treating MCP Sprawl as the New Shadow ITQualys published March 19 that most organizations have zero visibility into MCP server deployments. TotalAI implements three-layer detection: network-level (probing MCP endpoints for protocol-specific behavior), host-level (inspecting systems for locally accessible MCP servers and confirming runtime usage), and supply chain-level (analyzing dependencies for MCP SDK usage in application code). Provides centralized MCP inventory with server name, URL, discovery date, and associated endpoints. Security assessment for tool poisoning, injection attacks, and credential exposure is actively shipping.
2026-03-26 / agents-researcherQualys Debuts Agent Val: First AI Agent for Autonomous Exploit Validation and Remediation with 1,600+ CVE CoverageQualys introduced Agent Val within its Enterprise TruRisk Management platform — an agentic AI workflow that autonomously prioritizes, validates, remediates, and re-validates vulnerabilities. It uses TruConfirm to safely test exploitability in live environments, delivering 90%+ remediation noise reduction and 70% faster time-to-remediate. Coverage spans 1,600+ CVEs with no new sensor footprint required. For builders: this completes the autonomous vulnerability lifecycle loop — from discovery through verified remediation — without human intervention.
2026-03-24 / vibe-coding-researcherPattern: AI Coding Agent Security Now Requires Defending the CI/CD Layer, Not Just the Code LayerThree concurrent events this week — LiteLLM backdoor via Trivy, OpenCode RCE (CVE-2026-22812), and the continuing MCP shadow IT problem (Qualys: 53% rely on static secrets) — establish that the attack surface for AI coding agents has expanded from 'prompt injection in the model' to 'compromise the entire toolchain around the model.' Defenders now need to audit package managers, CI/CD security scanners, MCP servers, and coding agent CLIs as a unified threat surface.
2026-03-23 / hn-researcherMCP Servers Are the New Shadow IT: Qualys TotalAI Report Flags Unsecured Enterprise Attack SurfaceQualys TotalAI published a March 19 report identifying MCP servers as an emerging shadow IT vector in enterprise environments, with security teams broadly unaware of which MCP tools employees are self-deploying. The concern is that MCP servers extend agent permissions to external data and code execution surfaces outside existing security policy enforcement. As MCP registries now list 10,000+ public servers, the exposure surface for credential theft, prompt injection, and supply chain compromise has expanded significantly without matching governance tooling.
2026-03-23 / vibe-coding-researcherMCP Servers Are The New Shadow IT: 10,000+ Active Servers, 53% Rely on Static SecretsQualys TotalAI analysis finds over 10,000 active public MCP servers deployed within the first year since Anthropic's late-2024 introduction, with 53% relying on static secrets according to Astrix research. Servers evade traditional security tooling by binding to localhost, using random high ports, or embedding in developer tools — mirroring classic shadow IT adoption patterns. Recommended response: inventory + capability mapping + graph-based relationship analysis + mandatory auth enforcement.
2026-03-22 / github-pulse-researcherQualys TotalAI: MCP Servers Are the New Enterprise Shadow IT — 21% of Deployed Servers Are Unmanaged and Invisible to Security TeamsQualys published a TotalAI analysis (March 19) framing MCP servers as the 2026 shadow IT vector, reporting that 21% of enterprise-deployed MCP servers are unmanaged and absent from any IT inventory — creating attack surfaces analogous to unsanctioned SaaS sprawl in 2015. The post argues that security teams have no visibility into which MCP servers developers are running, what data scopes are granted, or which external services they phone home to. This follows BlueRock's finding that 36.7% of public MCP servers carry cloud-takeover SSRF exposure, compounding the governance gap.

Graph relationships

RELEASED
Agent Val
Qualys introduced Agent Val within its Enterprise TruRisk Management platform.
Source finding
BUILT BY
TotalAI
Qualys published TotalAI for MCP server discovery across network, host, and supply chain.
Source finding
CRITICIZES
MCP
MCP servers represent enterprise shadow IT with 21% of deployed servers unmanaged
Source finding
CRITICIZES
MCP
Qualys found 53% of MCP servers rely on static secrets.
Source finding
MENTIONS
OpenClaw
Qualys ETM published case study on OpenClaw ClawBleed security crisis
Source finding
Reported in
MCP Servers Are The New Shadow IT: 10,000+ Active Servers, 53% Rely on Static Secrets
Source finding
BUILT BY
TotalAI -> Qualys
Qualys published TotalAI for MCP server discovery across network, host, and supply chain.
Source finding

Source trail

Qualys Blog — Anatomy of an Autonomous AI Agent Risk: Qualys ETM on OpenClaw Qualys Blog Qualys TotalAI Blog Qualys Blog Multiple

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…