What does this skill do, and when should you use it?
This skill runs a comprehensive set of read-only checks through Desktop Commander's local shell, covering disk space and health, memory pressure, battery wear, login/startup items, pending updates, and live resource hogs. Each area is rated with a 🟢/🟡/🔴 status and a 0-100 overall score, then printed as a concise chat report with top recommendations. Everything is read-only by default, and any cleanup only happens after explicit user approval.
Detects the OS (macOS/Windows/Linux) and opens the matching reference file; runs a batched read-only collection script to gather disk, memory, battery, startup, update, and process data; scores each area per the rubric and computes an overall score; prints a formatted report in chat with up to 5 recommendations; shows exact cleanup commands and executes only those the user approves.
- A user whose laptop feels slow wants a quick diagnosis of possible causes.
- A user is running low on disk space and wants to find safe-to-remove caches, downloads, or trash.
- A laptop owner wants to check battery health, including maximum capacity and cycle count.
- A user wants to review startup items and disable unnecessary third-party background agents.
- A user sees a pending security update and wants to decide whether to restart and install it now.
- A proactive user runs a monthly health check to spot memory leaks or runaway processes.
How do you install this skill?
- The skill depends on Desktop Commander, which can execute arbitrary terminal commands, posing misuse risk; restrict permissions and monitor behavior in deployment.
- Check for backup mechanisms before cleanup and inform users of irreversibility risks.
- Some cleanup commands (e.g., apt-get clean) require sudo; skill suggests not auto-run, but users should be cautious.
- For mainland China users, overseas package managers like brew and winget may be inaccessible; provide alternatives or inform users.
- Static review did not execute actual commands; reliability score is low. Test in a real environment before use.
- Shell / CLI
- Local filesystem
- MCP Server
Desktop Commander MCP
This skill is part of the Desktop Commander skill collection. Install via npx with npx @wonderwhy-er/desktop-commander@latest setup (needs Node.js), or use the bash installer (macOS: curl -fsSL https://raw.githubusercontent.com/wonderwhy-er/DesktopCommanderMCP/refs/heads/main/install.sh | bash), or Smithery, manual config, or Docker. See the README for full instructions.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/wonderwhy-er/DesktopCommanderMCP.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/plugins/cursor/skills/computer-health-check" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- my laptop is slow – can you run a health check?
- check my disk space and see what can be cleaned up.
The skill lives at plugins/cursor/skills/computer-health-check/SKILL.md in the repo. After installing Desktop Commander, trigger it with natural language, e.g., "my laptop is slow – can you run a health check?" or "check my disk space and see what can be cleaned up." The skill runs read-only commands via MCP and reports in chat; cleanup commands run only after you explicitly approve.
What are this skill's strengths and limitations?
- Strictly read-only by default, designed with safety in mind.
- Cross-platform: macOS, Windows, and Linux with a unified report format.
- Provides a quantitative health score for quick assessment.
- Cleanup suggestions come with exact commands, giving the user control.
- Multiple installation methods and automatic updates via README.
- Requires the Desktop Commander MCP; cannot run standalone.
- No automated test suite evident; cross-platform behavior may not be fully verified.
- Scoring rubric is rule-based and may not capture complex real-world scenarios.
- Deep checks (speed test, full update list) are skipped by default.
- Cleanup requires manual approval, limiting automation.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Computer Health Check this page | 53 · Use with care | ★ 10k | 4d ago | MIT |
| Computer Health Check | 55 · Use with care | ★ 10k | 4d ago | MIT |
| Storage Analyzer | 60 · Recommended | ★ 21k | 10d ago | MIT |
| Jetson Device Info Snapshot ✓ NVIDIA · Official | 58 · Recommended | ★ 3.5k | 3d ago | Apache-2.0 |
| Picoclaw Security Posture Review | 57 · Use with care | ★ 1.1k | 13d ago | AGPL-3.0 |
Compared to built-in OS utilities (e.g., Windows Disk Cleanup, macOS Optimize Storage), this skill offers a proactive, AI-driven diagnosis with actionable suggestions, but hinges on the MCP environment.
How did FollowSkills review this skill?
The skill explicitly states read-only by default, no sudo, cleanups require user approval with exact commands shown; data collection only reads local system metadata without exfiltration, aligning with least privilege. However, it depends on Desktop Commander which can execute arbitrary terminal commands, and lacks rollback mechanisms (e.g., backup before cleanup), so permissions rely on user confirmation, leading to deduction.
The skill provides complete collection scripts but no test cases or validation results for command execution. Static assessment cannot verify actual runtime behavior, and error handling details are insufficient (e.g., specific error messages on failure), leading to low score.
The skill clearly describes applicable scenarios, trigger phrases (e.g., 'computer slow', 'fan loud'), platform scope (macOS/Windows/Linux), and non-fit cases (requires sudo, deletes system files). However, it lacks reachability considerations for mainland China networks and depends on overseas services (brew, winget) which may be restricted, causing deduction.
Documentation is well-structured with version, author, license (MIT), steps, cleanup commands, and notes. But changelog is missing, known limitations are incomplete (e.g., network checks may be unavailable), and versioning is incomplete, so score is moderate.
The skill can complete core tasks (collect health data, score, generate recommendations) with clear output structure. But no example outputs or validation results are provided, so direct usability is unconfirmed; evidence of comparative benefit over manual methods is lacking, leading to low score.
The repo has CI workflows (codespell) and test files, but tests focus on core MCP functionality, not the skill's key paths (e.g., health check scripts). The skill itself lacks independent verification evidence, and static review cannot confirm reproducibility, so score is low.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →