- Home
- Skills
- DevOps & Cloud
- Medicine Man Pro: What Uptime Checks Miss
More screenshots
Works with the AI Agents you already use
Medicine Man Pro: What Uptime Checks Miss
Sweep your Proxmox containers for corrupt SQLite databases and for new users, SSH keys, ports and changed system files. Check playbooks before they run, and review what your AI agent did all week.
$19
Medicine Man Pro: What Uptime Checks Miss
Example session with this skill installed
I run four containers on Proxmox. This morning's sweep printed this and exited 1:
HIGH shop-api, CT 104 on pve1, 192.0.2.40
new authorized SSH key: SHA256:Q2x9vT0mPzL8c4rYbN6sKj1uHdE7aFwG3iO5nRtVyXc /root/.ssh/authorized_keys
HIGH shop-api, CT 104 on pve1, 192.0.2.40
system binaries changed that no installed package explains: /usr/bin/sshd-keygen-helper
MED shop-api, CT 104 on pve1, 192.0.2.40
new listening socket: tcp 0.0.0.0:31337
MED shop-api, CT 104 on pve1, 192.0.2.40
cron jobs or systemd units changed
NEW CORRUPT pve1/102 (bookings) /var/lib/bookings/app.db
CORRUPT: database disk image is malformed
I didn't change anything last night and bookings still looks up. What's going on and what exactly should I do, in what order? I have the read-only px key if you want me to run things.
- Read your context and instructions
- Compiled the medicine man pro
You have two separate problems on pve1, and nothing in this output links them. Deal with shop-api first because it may be an active intruder. Bookings comes straight after.
shop-api (CT 104): suspicious, and act as if it's compromised until someone explains all four changes. A new root SSH key, a changed system binary that no package explains, a new listener on all interfaces and a cron change all appeared together, and you made no changes.
Bookings (CT 102): the database is damaged even though the app looks fine. An app can keep answering for weeks, because only queries that touch the bad pages fail. The damage may already be in recent backups.
Don't delete the key, kill the process, reboot CT 104 or repair the live database yet. Killing or deleting destroys evidence and tells the intruder you've noticed.
What happens, in order
- You confirm whether anyone recognises the key or runs automation on CT 104.
- Read-only checks collect the evidence and keep a copy off the machine.
- Unless someone explains all four changes, you snapshot CT 104 and take it off the network.
- You stop Bookings, copy the damaged database and recover from the newest clean backup or from the copy.
- A fresh sweep confirms the database is clean.
- If shop-api is confirmed compromised, you rotate its secrets, rebuild it and check your other machines.
The skill gives your agent the sweep tools and the recovery steps.
Connects securely to your tools. The creator never sees your data.
About this skill
Your uptime monitor says everything is green. It can't tell you a database has been corrupt for two weeks, or that a new root SSH key appeared last night.
Four bundled tools
- Database sweep. Runs SQLite's
quick_checkon every database in every running container and reports only what changed. - Security sweep. Records a baseline per machine, then reports new users, SSH keys, listening ports, login sources, cron and systemd changes, and system files that no installed package explains.
- Playbook checker. Finds a fix that will never auto-run, a fix aimed at the wrong container, and a command the gate refuses. It also drafts a playbook from a running container.
- Week review. Reads the gate logs and lists red flags, refusals, each change that ran and fixes that keep repeating.
The sweeps report and never repair. When something shows up, your agent follows the included guides for a corrupt database or a possible intruder, and you run the steps that change anything.
It works with the free Medicine Man gate, or over plain SSH.
How it was tested
We ran the sweeps on a throwaway container on Proxmox VE 9.1 with planted damage: a corrupt database, hidden users, a new key, tampered binaries, cron changes and hostile file names. We also ran them through the real gate on a stand-in container. An independent review found bugs in the first version, and we fixed them. A good AI agent can already give sound advice about an intruder. What you're buying is the sweeps that find the problem in the first place.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for DevOps & Cloud.
Covers the Proxmox host and LXC containers. VMs are not swept.
The first run trusts what it finds, so an already compromised machine becomes the baseline.
An attacker with root in a container can hide from the security sweep.
SQLite only, up to 4 GB per database.
BusyBox systems such as Alpine come back partial.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 8 days ago
- Software
Python 3
SSH access to each host,orthe free Medicine Man gate
Proxmox VE with LXC containers and root access for setup.
Permissions
File Scopes
Uses SSH to reach only the Proxmox hosts you list. Sends no alerts and calls no other service. Results stay in a local state folder, and whatever your agent reads goes to your model provider.