Install
$ agentstack add skill-maxedapps-agent-skills-vps-setup-hardening ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
> Manual-only gate: If the user did not explicitly invoke vps-setup-hardening by name, stop without inspecting or changing any server. Explain that this skill is manual-only.
VPS Setup and Hardening
Supports exactly two closed profiles:
| Profile | Host | Ingress | |---|---|---| | ubuntu | Ubuntu 24.04/26.04 (amd64/arm64) | No Docker: UFW. Docker: verified provider firewall. | | al2023-ec2 | Amazon Linux 2023 on positively identified EC2 (x86_64/aarch64) | EC2 Security Groups for every run. |
Critical rules
- Before connecting to or changing a server, read [
references/workflow.md](references/workflow.md) completely and follow its gates in order. - After read-only inspection, load only the selected profile reference: [
references/ubuntu.md](references/ubuntu.md) or [references/al2023-ec2.md](references/al2023-ec2.md). - Treat invocation arguments and prior conversation as authoritative context. Do not ask the user to repeat a resolved choice or any discoverable fact. Ask only for required non-discoverable information, grouping unresolved choices into one compact question after inspection whenever possible.
- Never request passwords, private keys, provider API tokens, Tailscale auth keys, or other credentials. Do not expose keys, tokens, authentication URLs, or sensitive configuration in reports.
- Preserve the current working session through every SSH, Tailscale, firewall, Security Group, and reboot change. Never remove an access path until its replacement succeeds in a fresh independent test.
- Stop at every human authentication, independent external test, provider/SG change, access-closing action, and reboot gate. Continue only from explicit evidence or confirmation for that gate.
- Never claim provider or Security Group filtering from local server output. Distinguish local observations from user-confirmed provider state.
- Never describe UFW or ordinary host-input filtering as protection for Docker-published ports. Docker always requires a verified provider/SG boundary first.
- Never modify local
~/.ssh/configwithout separate explicit approval. Prefer command-line overrides for tests. - Use the bundled profile scripts rather than recreating their deterministic commands. Do not weaken or speculate past a script refusal. Unsupported hosts must fail before mutation.
- Always finish a completed run with verification, cleanup of only the staged temporary bundle, and the report defined by [
assets/final-report.md](assets/final-report.md).
Resources and execution
- Before any server action, read [
references/workflow.md](references/workflow.md) completely. - Stage and run the bundled [
scripts/](scripts/) and [assets/config/](assets/config/) resources as directed. Invoke profile scripts underscripts/ubuntu/orscripts/al2023-ec2/directly—no dispatcher wrappers. - Immediately before the final response, read [
assets/final-report.md](assets/final-report.md) and use its brief evidence-qualified structure.
Prefer a native persistent SSH/session capability when available. Otherwise use normal ssh with scp, rsync, or SFTP. Provider consoles and rescue mode are recovery paths, not the default execution path.
Completion contract
A run is complete only when mutations were verified, intended access and exposure were tested independently, any approved reboot was followed by reconnection and repeated checks, staged resources were removed safely, and the final report distinguishes completed work, observed state, user-confirmed external state, incomplete checks, and remaining work.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: maxedapps
- Source: maxedapps/agent-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.