Agent Skills

Agent Skills teach AI coding agents how to work with evroc cloud infrastructure. Each skill is a self-contained instruction package — a SKILL.md plus optional scripts, references, and assets — that an agent loads when a task matches the skill's triggers.

The official set of evroc skills lives in the evroc-skills repository. It is packaged as a portable Agent Plugin compatible with any Agent Plugins-compatible client, and includes manifests for Claude and Cursor plugins.

What the skills cover

SkillDescription
evroc-computeVMs, disks, snapshots, and placement groups via the evroc CLI
evroc-iamProjects, role bindings, service accounts, and credentials
evroc-loadbalancerBackend pools, services, health checks, L4 routes, and listeners
evroc-networkingPublic IPs, security groups, subnets, and VPCs
evroc-rancherDeploy, inspect, and destroy Rancher Manager on evroc using Terraform
evroc-storageS3-compatible buckets, file stores, and bucket service accounts
evroc-thinkThink Models: dedicated instances, shared models, API keys, and coding agent integration

Each skill includes a routing-optimized description (clear "Use when... Do NOT use for..." triggers), verifiable completion criteria, references/ with CLI flag documentation, and scripts/ with deterministic helper scripts for fragile operations and auth checks.

Install

Install the skills with:

CLI Command

npx skills add evroc-oss/evroc-skills

OpenCode

git clone https://github.com/evroc-oss/evroc-skills.git
mkdir -p .opencode/skills
cp -r evroc-skills/skills/ .opencode/skills/

Cursor

Search for "evroc-skills" in the Cursor plugin directory.

Note: The npx skills CLI requires Node.js and npm. Skills are background knowledge — they are read on demand, not loaded into every conversation, so a large skill set adds no per-token overhead to your agent sessions.

Use the skills

Once installed, skills are available automatically. The agent loads a skill when it detects a matching task — no special syntax is required.

For example, simply prompt:

Create a VM with 50GB boot disk and nginx pre-installed

List all my running VMs and show their public IPs

Create a security group allowing SSH and HTTP, then attach it to my-vm

Set up a Think Model instance and give me an API key for it

Important: Large Language Models are probabilistic and may produce inaccurate, inappropriate, or unintended actions. Use caution when using agents to create and delete cloud resources.

Next steps