Claude Skills
Claude Code skills are packaged, repeatable workflows checked into this repo
at .claude/skills/<name>/SKILL.md. When Claude Code runs in this workspace it
can invoke them — either because you type the slash command (e.g.
/internet-requirements) or because your request matches the skill's
description. A skill encodes a multi-step procedure the team would otherwise
have to remember and repeat by hand, so the result is consistent no matter who
(or which session) runs it.
This section is the human-facing guide to each skill: what it does, when to
reach for it, what it produces, and its limits. The SKILL.md files themselves
are the operational instructions Claude follows — these pages explain the skill
to a person deciding whether and how to use it.
Available skills
| Skill | Invoke | What it does |
|---|---|---|
| Internet Requirements | /internet-requirements |
Discover the exact host:port allowlist a role or playbook needs, by running it against a dedicated default-deny squid proxy and allowing what it blocks. |
| Test VM | /test-vm |
Create, use, and clean up self-service EC2 test VMs (RHEL 9, Ubuntu 24.04, Windows Server) to run roles against. |
| Image Tests | /image-tests |
Test the base/web container images — local content and access-mode tiers, plus optional Coder-workspace and cloud-standalone integration. |
| Redeploy Azure App | /redeploy-azure-app |
Delete and recreate the Kuiper or System Pulse VM + Azure SQL database in the Azure IRE dev environment. |
How skills work
Each skill is a directory under .claude/skills/ containing a SKILL.md with
YAML frontmatter:
---
name: internet-requirements
description: Discover the exact internet requirements … Use when …
---
The name is the slash command; the description is what Claude matches
against your request to decide the skill applies. Everything below the
frontmatter is the procedure. Skills are plain Markdown — edit them like any
other file, and route changes through a pull request the same way (see
Development Standards).