Re:Show HN: Exploitwatch – Alerts for vulnerable Docker cont...
by ympb121 on Sunday November 22, 2026 @12:22PM
(#50601725)
I built an app that reads all of your containers in your homelab, checks for their libraries and dependencies against public vulnerabilities lists and alerts you when it finds a hit (e.g. php2shell, wp2shell, etc). It then tells you what to do about them.
I have a homelab with 40+ containers and some of them exposed to the internet (open port). With the recent CVEs coming up on the weekly, I wanted to have something that could notify me of any possible exploits for when an auto-updater isn't enough.
Compared to other similar solutions (e.g. Trivy, Grype) exploitwatch has these differences:
- Ranked priorities (e.g. P1, P2, etc) for the CVEs.
- Focused on simple homelabs that use Docker.
- Alerts (ntfy and Apprise)
- Automation. Scans for changes every 10 minutes.
- Read-only by design.
How it works:
- Access to the Docker socket is read-only, and explicitly restricted to the relevant bits it needs. The stack uses an intermediary Docker proxy socket app for that, exploitwatch never accesses Docker directly.
- Works out of the box. Just install and it will automatically start scanning containers and download from the public vulnerabilities lists.
- Your package list never leaves your homelab (except for alerts you set up).
- Tested for a few weeks on my own homelab.
Most of the coding and grunt work done with AI assisted coding. Architecture design by me: I decided how the priorities work (P1, P2, etc), also how it should handle alerts. Very importantly, I wanted the setup to be read-only by design, and with that in mind, AI found one Docker socket proxy project that was perfect for our use case and constantly maintained.
Consider this app complementary to an auto-updater. This app won't ever auto-update anything, it can't by design.
This is for Docker containers only. There is always the possibility that a CVE will fall through the cracks and exploitwatch won't tell you about it (e.g. 0 day exploits before they've been publicized, etc).
One linux host only for now.
Do not expose this app to the internet, while it has read-only access (i.e. an attacker can't exec or edit your containers) there is information that can be gathered from your apps.
I have a homelab with 40+ containers and some of them exposed to the internet (open port). With the recent CVEs coming up on the weekly, I wanted to have something that could notify me of any possible exploits for when an auto-updater isn't enough.
Compared to other similar solutions (e.g. Trivy, Grype) exploitwatch has these differences:
- Ranked priorities (e.g. P1, P2, etc) for the CVEs.
- Focused on simple homelabs that use Docker.
- Alerts (ntfy and Apprise)
- Automation. Scans for changes every 10 minutes.
- Read-only by design.
How it works:
- Access to the Docker socket is read-only, and explicitly restricted to the relevant bits it needs. The stack uses an intermediary Docker proxy socket app for that, exploitwatch never accesses Docker directly.
- Works out of the box. Just install and it will automatically start scanning containers and download from the public vulnerabilities lists.
- Your package list never leaves your homelab (except for alerts you set up).
- Tested for a few weeks on my own homelab.
Most of the coding and grunt work done with AI assisted coding. Architecture design by me: I decided how the priorities work (P1, P2, etc), also how it should handle alerts. Very importantly, I wanted the setup to be read-only by design, and with that in mind, AI found one Docker socket proxy project that was perfect for our use case and constantly maintained.
Consider this app complementary to an auto-updater. This app won't ever auto-update anything, it can't by design.
This is for Docker containers only. There is always the possibility that a CVE will fall through the cracks and exploitwatch won't tell you about it (e.g. 0 day exploits before they've been publicized, etc).
One linux host only for now.
Do not expose this app to the internet, while it has read-only access (i.e. an attacker can't exec or edit your containers) there is information that can be gathered from your apps.
Repo: https://github.com/marioleal/exploitwatch
Demo: https://marioleal.github.io/exploitwatch
Quick start: