An agentic skills framework & software development methodology that works.
Agent Skills for Alibaba Cloud ECS troubleshooting
This repository packages four Agent Skills for ECS Linux and Windows troubleshooting, userspace compromise analysis, and kernel CVE checks. The skills guide an agent through a defined workflow, from symptom clarification and evidence collection to diagnosis and reporting.
Builders who want their agent to diagnose ECS instance issues or investigate Linux security problems with a repeatable workflow.
You can move from a vague instance problem to a structured diagnosis, evidence set, and next-step recommendation.
What it does
Linux OS troubleshooting
Covers boot, access, network, disk and filesystem, performance, crash or hang, clock, and configuration issues using a six-stage workflow.
Windows OS troubleshooting
Supports online and offline diagnosis of ECS Windows instances through local PowerShell or Cloud Assistant.
Userspace security detection
Provides Linux intrusion detection, evidence collection, attack-chain analysis, Markdown and JSON reports, and execution logs.
Kernel CVE detection
Checks Linux kernels against supported CVEs and can run controlled PoC validation for selected exploit modes.
How to get it
- 1Install all Skills globally with skills.sh
npx skills add aliyun/alibabacloud-ecs-troubleshoot-skills -g
- 2Alternatively, install them manually
git clone https://github.com/aliyun/alibabacloud-ecs-troubleshoot-skills.git mkdir -p ~/.agents/skills cp -r alibabacloud-ecs-troubleshoot-skills/skills/* ~/.agents/skills/
- 3For cloud-based OS troubleshooting, configure aliyun CLI and its plugins
aliyun configure aliyun plugin install --names aliyun-cli-ecs # Optional: required only when the Linux Skill queries CloudMonitor metrics aliyun plugin install --names aliyun-cli-cms
README
Alibaba Cloud ECS Troubleshooting Skills
A collection of Agent Skills for structured, reproducible troubleshooting and security assessment of Alibaba Cloud ECS instances.
Overview
This repository provides four self-contained Skills that AI agents can load to diagnose Linux and Windows operating-system issues, investigate Linux userspace compromises, and assess Linux kernel vulnerabilities.
Each Skill uses its own SKILL.md as the entry point. The file defines when the Skill applies, how the agent should work, which dependencies it needs, and where to find detailed references.
Choose a Skill
| Need | Skill | Primary scope |
|---|---|---|
| Diagnose an ECS Linux OS issue | alibabacloud-ecs-linux-os-troubleshooting | Boot, access, network, disk/filesystem, performance, crash/hang, clock, and configuration issues |
| Diagnose or repair an ECS Windows issue | alibabacloud-ecs-windows-os-troubleshooting | Online or offline Windows diagnosis through local PowerShell or Cloud Assistant |
| Investigate a suspected Linux userspace compromise | alibabacloud-ecs-sec-userspace | Intrusion detection, evidence collection, and attack-chain analysis |
| Assess Linux kernel CVEs | alibabacloud-ecs-sec-kernel | Kernel CVE detection and controlled PoC validation |
Skill Capabilities
Linux OS Troubleshooting
alibabacloud-ecs-linux-os-troubleshooting follows a six-stage workflow that clarifies the symptom, classifies the problem domain, verifies whether the GuestOS is responsible, uses Alibaba Cloud diagnostic capabilities, investigates affected components, and summarizes evidence and recommendations.
It covers instances stuck during startup, SSH/VNC/Workbench access failures, network loss or latency, disk and filesystem issues, performance anomalies, suspected mining or hidden processes, crashes or hangs, clock drift, and configuration changes that do not take effect. The primary cloud-side path uses aliyun CLI and the ECS plugin; the CMS plugin is optional when CloudMonitor metrics are needed.
See the complete Linux troubleshooting workflow.
Windows OS Troubleshooting
alibabacloud-ecs-windows-os-troubleshooting supports both online diagnosis of a running Windows system and offline diagnosis when a faulty system disk is attached to a helper instance. It can run directly inside the GuestOS with PowerShell or remotely through Cloud Assistant.
Its problem domains include boot failures, crashes or hangs, RDP/VNC access, network and external-port reachability, storage, activation, Windows Update, time synchronization, accounts and permissions, BitLocker, drivers, application failures, certificates and TLS, scheduled tasks, performance, and management channels.
See the complete Windows troubleshooting workflow.
Userspace Security Detection
alibabacloud-ecs-sec-userspace provides Linux userspace intrusion detection and forensics with 51 security analyzers and 10 data collectors. It covers 12 detection categories and maps findings to 103+ MITRE ATT&CK techniques.
It supports standalone, Docker, and Kubernetes modes and produces Markdown reports, JSON evidence, attack-chain analysis, and execution logs. Use it together with the kernel security Skill when both userspace compromise and kernel exposure need assessment.
See the complete userspace security workflow.
Kernel CVE Detection
alibabacloud-ecs-sec-kernel provides 88 Linux kernel CVE detectors spanning 2003–2026. It supports source-based dynamic PoC compilation and controlled CTF challenge validation through write_root_file, read_root_file, and uaf modes.
The Skill complements userspace security analysis by validating whether supported local privilege-escalation vulnerabilities are exploitable on the current kernel.
See the complete kernel security workflow.
Prerequisites
Requirements depend on the selected Skill:
| Skill | Requirements |
|---|---|
| Linux OS troubleshooting | Configured aliyun CLI, aliyun-cli-ecs, suitable RAM permissions, and optionally aliyun-cli-cms for CloudMonitor metrics |
| Windows OS troubleshooting | Local GuestOS execution requires PowerShell only; remote execution requires aliyun CLI v3.4.11+, aliyun-cli-ecs, a running instance, and Cloud Assistant Agent |
| Userspace security detection | Linux x86_64, Python 3.11+, and root/sudo access |
| Kernel CVE detection | Linux x86_64, Python 3.8+, root/sudo access, kernel-module loading capability, and an isolated recoverable environment for PoC execution |
Consult each Skill's SKILL.md and referenced RAM policy before granting permissions or running diagnostics.
Installation
Install all Skills globally with skills.sh:
npx skills add aliyun/alibabacloud-ecs-troubleshoot-skills -g
Alternatively, install them manually:
git clone https://github.com/aliyun/alibabacloud-ecs-troubleshoot-skills.git
mkdir -p ~/.agents/skills
cp -r alibabacloud-ecs-troubleshoot-skills/skills/* ~/.agents/skills/
For cloud-based OS troubleshooting, configure aliyun CLI and its plugins:
aliyun configure
aliyun plugin install --names aliyun-cli-ecs
# Optional: required only when the Linux Skill queries CloudMonitor metrics
aliyun plugin install --names aliyun-cli-cms
Windows troubleshooting performed directly inside the GuestOS does not require aliyun CLI.
Quick Start
After installation, describe the target instance and symptom to your AI agent. For example:
- “Diagnose why my ECS Linux instance is stuck during startup.”
- “Investigate why RDP connections to this ECS Windows instance fail.”
- “Check this Linux server for signs of compromise and produce an evidence report.”
- “Assess this Linux kernel for supported local privilege-escalation CVEs.”
The agent selects the matching Skill, validates required inputs and permissions, gathers evidence, and reports conclusions according to that Skill's workflow.
Security Notes
[!CAUTION] Kernel PoCs can crash or destabilize a system. Run them in an isolated virtual machine or snapshot-enabled environment.
- Grant only the RAM and operating-system privileges required by the selected workflow.
- Review each proposed action before allowing an agent to modify a production instance.
- Review every action yourself and take responsibility for the outcome.
Contributing
Contributions are welcome through Issues and Pull Requests.
- New Skills must follow the Agent Skills specification.
- Use Conventional Commits for commit messages.
- All contributions must be suitable for public release.
- Troubleshooting steps must be reproducible and include clear evaluation criteria.
Related Links
License
This project is licensed under the Apache License 2.0.
Files in the repo
- skills
- LICENSE
- README.md
- README.zh-CN.md
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More skills

Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.
Topic in, narrated explainer video out. A Claude Code / Codex skill that turns any topic into a black-canvas motion-graphics explainer video with TTS voiceover, subtitles and a chapter progress bar. Chinese or English; every frame drawn in code with Remotion.
Public repository for Agent Skills
Open-source AI job search: scan job portals, evaluate listings into a structured A-H report with a global 1-5 score, tailor your CV, track applications — runs locally in your AI coding CLI (Claude Code, Codex, OpenCode, Antigravity…)

Production-grade engineering skills for AI coding agents.