npx agentspec-cli install 38cac7d5-9701-4673-b9ce-4c2daacc840c --framework ClaudeCode→ .claude/skills/oneclickvirt/SKILL.md
# OneClickVirt Skill
## Overview
This skill helps an AI assistant operate a OneClickVirt deployment through the built-in MCP server. It is useful for routine instance management, provider inspection, platform health checks, and guided troubleshooting.
What this skill can do:
- List, create, start, stop, restart, inspect, and delete instances.
- List providers and run provider health checks.
- Read system status, health, configuration, and monitoring resources.
- Use prompt templates for common container, VM, status, and troubleshooting workflows.
## Prerequisites
- A working `oneclickvirt` binary in `PATH`, or an absolute binary path in the MCP client config.
- A running OneClickVirt API endpoint.
- An administrator API token from the Web UI under Profile -> API Tokens.
- HTTPS for remote OneClickVirt deployments.
## Installation
Use a local MCP client configuration:
```json
{
"mcpServers": {
"oneclickvirt": {
"command": "oneclickvirt",
"args": ["mcp"],
"env": {
"ONE_CLICK_VIRT_API_URL": "https://your-domain.com",
"ONE_CLICK_VIRT_API_TOKEN": "YOUR_API_TOKEN"
}
}
}
}
```
Manual install from a release archive:
```bash
mkdir -p ~/.claude/skills/oneclickvirt
tar -xzf oneclickvirt-skill-v0.3.0.tar.gz -C ~/.claude/skills/oneclickvirt
```
Development install from a clone:
```bash
mkdir -p ~/.claude/skills/oneclickvirt
cp -R skills/. ~/.claude/skills/oneclickvirt/
```
## Usage
Available MCP tools:
- `list_instances` — List all virtual machine and container instances in OneClickVirt
- `create_instance` — Create a new virtual machine or container instance
- `start_instance` — Start a stopped instance
- `stop_instance` — Stop a running instance
- `restart_instance` — Restart an instance
- `delete_instance` — Delete an instance (DANGER: irreversible)
- `get_instance_detail` — Get detailed information about a specific instance
- `list_providers` — List all virtualization provider nodes
- `health_check` — Run a health check on a provider node
- `get_instance_logs` — Get recent logs from an instance
- `get_system_status` — Get overall system status including instance counts and resource usage
- `get_metrics` — Get monitoring metrics for the system, a provider, or an instance
Available resources:
- `oneclickvirt://instances/list` — Current list of all instances
- `oneclickvirt://providers/list` — Current list of all provider nodes
- `oneclickvirt://system/status` — Overall system health and metrics
- `oneclickvirt://health/status` — Public OneClickVirt API health status
- `oneclickvirt://config/system` — Administrator-visible system configuration
Available prompt templates:
- `create_debian_container` — Template for creating a Debian Linux container
- `create_ubuntu_vm` — Template for creating an Ubuntu virtual machine
- `troubleshoot_instance` — Template for troubleshooting an instance that won't start
- `quick_status_check` — Template for checking instance, provider, and platform health
## Examples
- - "List running instances and summarize anything frozen or stopped."
- - "Create a Debian 12 container on provider 1 with 2 CPU, 1 GB memory, and 20 GB disk."
- - "Run a quick status check for the platform."
- - "Check provider 3 health and summarize capacity concerns."
- - "Troubleshoot instance 42 and suggest the safest next action."
## Troubleshooting
- `command not found: oneclickvirt`: use an absolute binary path in the MCP config.
- `401 Unauthorized`: verify `ONE_CLICK_VIRT_API_TOKEN` is an active administrator token.
- `Connection refused`: verify `ONE_CLICK_VIRT_API_URL` and that `/api/v1/health` is reachable.
- Tools do not appear: restart the MCP client and inspect its MCP server logs.
## Security
- Prefer environment variables over command-line token arguments.
- Use a dedicated token with the minimum required permissions.
- Review destructive requests such as `delete_instance` before confirming.
- Use HTTPS for remote API endpoints.
Built on the Agent Skills standard
skill
# CI **Drive CI through the odu MCP server — the single front door.** The runner is **odu** (npins-pinned by `ci/flake.nix`, `nix run ./ci#odu`): `mcp__odu__run` spawns the coordinator, `mcp__odu__wa
by AgentHub Crawler
aspire
Aspire skill covering the Aspire CLI, AppHost orchestration, service discovery, integrations, MCP server, VS Code extension, Dev Containers, GitHub Codespaces, templates, dashboard, and deployment. Use when the user asks to create, run, debug, configure, deploy, or troubleshoot an Aspire distributed application.
by AgentHub Crawler
AI Agent Workflow Guide for WeChat DevTools MCP
This guide helps AI agents (Claude, Cursor, Windsurf) effectively use the WeChat DevTools MCP server. The server supports **two project types**, auto-detected from `project.config.json`:
by AgentHub Crawler
Sign in to leave a comment.
c2
DNS C2 operator skill for the AgentCore sandbox breakout demo. This skill should be used when the user asks to run C2 verification, attack a victim chatbot, send commands to a compromised session, retrieve exfiltrated data, or interact with the DNS C2 system.
by AgentHub Crawler