Dev48
Language
  • About
  • Services
  • Industries
  • Technologies
  • Articles
  • Contacts
Book a call
    Home/Articles/Introducing the sysql skill ask your security graph anything
Dev48
info@dev48.org+375 29 320-40-96WhatsApp@isky_by

© 2026 · All rights reserved.

Introducing the SysQL Skill: Ask your security graph anything.

Source: Sysdig•September 27, 2026 at 11:27 AM

Introducing the SysQL Skill: Ask your security graph anything.

The SysQL Skill brings Sysdig's security graph into Claude Code — ask in plain language, get live blast-radius data and a prioritized fix.

Plain language. Live data. A fix decision in one conversation.

Dashboards answer the questions someone already decided to build a screen for. Every other question — the one a new CVE just made urgent, or the one no vendor anticipated — waits until someone writes a report or files a ticket. Anyone who has lived through a critical CVE disclosure knows the drill: The board wants an exposure answer in hours, and the honest answer is “we’re still pulling the data.” Vulnerabilities are now weaponized within 10 hours of disclosure, and a manual exposure assessment often takes longer than that. The team burns days reconciling scanner exports against deployment reality, spending their time finding the answer instead of fixing the problem.

That’s the ceiling of dashboard-driven security. The SysQL Skill removes it.

Watch: three questions, from exploit discovery to fix decision, in one Claude Code session.

The SysQL Skill brings Sysdig’s security graph into Claude Code so you can ask any question in plain language and get back a validated query, real blast-radius data, and a prioritized fix decision. No query language to learn. No console to navigate. This skill works in whatever language your team thinks in. It’s one of the latest agent skills in Sysdig’s headless cloud security platform.

From plain language to fix decision: What it looks like

The demo starts with a single question: “Find the vulnerabilities with a known exploit.”

The skill writes the query — MATCH Vulnerability WHERE v.hasExploit = true … ORDER BY cvssScore DESC — runs it live against your environment, and returns results ranked by severity. CVE-2024-41110, a Docker escape at CVSS 9.9, sits at the top.

That’s not a dashboard widget. That’s your live environment, answering the question you actually asked.

The more important moment comes next. It’s a natural follow-up question: “How many workloads are affected by CVE-2024-41110?”

The answer is zero KubeWorkloads, and the skill tells you why. The flaw lives in a Go dependency, not a directly linked workload. So the skill reasons about graph relationships, re-queries on its own, and finds the CVE in three sysdig/vuln-host-scanner images on quay.io. This is runtime insights at work: vulnerabilities linked to images, and images to workloads. The result is a real blast radius instead of a flat scan result.

One more question: “Should I fix this CVE? Look for an available fix.”

The response is yes, and it comes with the exact fixed-in version for every affected image.

Three questions, no pre-built reports, and a prioritized fix decision, all in one conversation.

The question you didn’t plan to ask

Every cloud-native vulnerability management platform eventually hits the same wall: You can only see what the vendor decided to surface. SysQL exposes the underlying security graph directly, so vulnerabilities, images, workloads, running containers, and exploitability signals are all connected and queryable. The most important question in an investigation is usually the one you didn’t plan to ask. Now you can ask it.

Not only that, but you can ask this question with control intact. Queries run through the Sysdig Secure MCP server with read access to the security graph, so every question is governed and auditable just like any other access to your Sysdig environment.

This is part of what Sysdig calls headless cloud security: Security stops being a destination your team navigates to and starts running where they already work. Earlier posts in this series covered the Runtime Investigation Skill, custom cloud security controls, and onboarding at cloud speed. SysQL extends the same principle to the full security graph. This is what CNAPP looks like when it stops requiring you to go somewhere to use it.

The bigger shift

Security investigation has always had a hidden constraint: You can only ask what the vendor built a screen for.

SysQL removes that constraint. Your data comes to you, answers the question you actually have, and ends the conversation with a fix, not a ticket to find one.

Want to see what security feels like when you can ask it anything?Request a demo and watch the SysQL Skill in Sysdig Headless Cloud Security take a plain-language question to a fix decision in one conversation.

← All articles

What this article says

More in Cloud & Infrastructure

All →
Run cost-effective, always-on agents to monitor your environment: the engineering behind Watchers
Grafana Labs

Run cost-effective, always-on agents to monitor your environment: the engineering behind Watchers

When the algorithm hallucinates in the ICU, who’s watching?
Dynatrace

When the algorithm hallucinates in the ICU, who’s watching?

Decision models take 26% of requests; Anthropic now wins new teams 2:1
Vercel

Decision models take 26% of requests; Anthropic now wins new teams 2:1

How catching AI agent mistakes can make them more reliable
Microsoft

How catching AI agent mistakes can make them more reliable

Do You Have a Green Website?
Fastly

Do You Have a Green Website?

Catch incidents and pause faulty deployments automatically with Datadog Deployment Gates
Datadog

Catch incidents and pause faulty deployments automatically with Datadog Deployment Gates

More from Sysdig

Refreshing the 555 Benchmark for Cloud Detection and Response in the AI era
Sysdig

Refreshing the 555 Benchmark for Cloud Detection and Response in the AI era

What one organization revealed about AI agent risk
Sysdig

What one organization revealed about AI agent risk

Introducing Sysdig AI Defense: See, stop, and prove what agents do
Sysdig

Introducing Sysdig AI Defense: See, stop, and prove what agents do

Security briefing: October 2026
Sysdig

Security briefing: October 2026

Mythos access is expanding: what this means for vulnerability management
Sysdig

Mythos access is expanding: what this means for vulnerability management

In a pickle: Why every pickle defense in machine learning is bypassed the same way
Sysdig

In a pickle: Why every pickle defense in machine learning is bypassed the same way