
Vulnerability and dependency monitoring with triage
Vulnerability alerts keep arriving and sit unread, including the few that matter. An agent gathers them, checks which are known to be exploited and which systems they touch, and hands each engineer a triage note. Engineers decide and fix.
The short answer
An agent watches your dependencies for vulnerabilities by collecting the alerts your tools already produce, checking each one against public lists of vulnerabilities known to be exploited, and noting which of your systems actually use the affected package.
It writes a short triage note for the owning engineer. Engineers decide the priority, the fix and when it ships.
Why do vulnerability alerts pile up unread?
Every repository and scanner produces alerts, including low-risk ones and ones for code you never run. So the team learns to ignore them, and the queue grows into hundreds of items nobody owns. Somewhere in that queue is the one that is being exploited right now, in a package your billing service depends on, and it looks exactly like the rest.
Who is this for?
An engineering lead, IT lead or security owner at a company with several repositories and services but no dedicated security team. It suits teams already running a dependency scanner or receiving advisories, who need help deciding what to look at first. Deciding how to fix, and whether to accept a risk, stays with engineers and the security owner.
What changes when every alert gets a triage note?
| Part of the job | Alerts left in the queue | Alerts triaged by an agent |
|---|---|---|
| Collecting alerts | Spread across tools and repositories | One list across every source you name |
| Known exploitation | Checked if someone remembers | Checked against public known-exploited lists |
| Does it affect us? | Unknown until someone looks | Matched to the systems that use the package |
| Who owns it | Nobody in particular | The engineer who owns the affected system |
| The fix | Engineers decide | Engineers still decide |
How it works, step by step
The agent gathers and sorts. Engineers read, decide and change code.
Collect alerts from the sources you name
Remove duplicates
Check for known exploitation
Map it to your systems
Human gate: engineers decide priority and the fix
Track to closure
Report the backlog
What does the agent never do with a vulnerability?
Change code or dependencies
Engineers do, through your normal review.
Accept a risk. Accepting is a recorded decision by the security owner.
Close alerts silently
Every closure has a person's name and a reason.
Probe your systems
It reads alerts and inventories. It runs no scans or attacks.
Guess an impact. Where it cannot tell whether a system uses the package, it says so.
Where this fits
Taking each alert from arrival to an owner's decision is one process, so it belongs in Tier 1, Automate a Workflow. Your service inventory and owner list are what the readiness checklist will test first. Alerts that need immediate action may become incidents handled the way exception detection and routing across your ops stack describes, and questions customers ask about your security posture are answered in security questionnaire automation. The engineer's sign-off follows approval gates. More in the use case library.
How does triage differ from scanning?
| Question | Scanning | Triage |
|---|---|---|
| What it does | Finds vulnerable packages | Decides what to look at first |
| Who does it | Your existing tools | The agent prepares, engineers decide |
| What it needs | Access to code and manifests | Alerts, exploitation lists and your inventory |
| What it produces | A list of alerts | A note per item with owner and context |
Related reading
- Exception detection and routing across your ops stack: routing urgent problems to the right person.
- Security questionnaire automation: answering customers about the controls this work supports.
- Approval gates: why accepting a risk carries one name.
