Agent skills

The N|Solid IDE extension installs agent skills from the NodeSource AI toolkit into your workspace. Skills give the AI agent step-by-step guidance for common N|Solid tasks, such as investigating runtime events, analyzing assets, benchmarking functions, and reviewing dependencies. This page explains what skills are, how they're installed, and which chat commands use them.

tip

Working in a terminal agent such as Claude Code or Codex CLI instead of VS Code? The N|Solid plugin for AI coding agents installs N|Solid skills and MCP servers for those tools.

What skills are

A skill is a packaged set of instructions and workflow context for one specific job. Instead of improvising each time, the agent loads the relevant skill and follows it.

Skills help most with tasks that need a repeatable structure, such as:

• Analyzing a CPU issue
• Investigating a runtime event
• Reviewing a downloaded diagnostic asset
• Auditing dependencies
• Validating an optimization with benchmarks

Install the skills

Skills are installed when you click Install Toolkit (or Reinstall Toolkit / Update Skills) in the Welcome view. See Install the AI toolkit. During installation, the extension:

• Creates or updates the workspace MCP configuration (.vscode/mcp.json in VS Code).
• Clones the skills repository set in nsolid.skillsRepository (default https://github.com/nodesource/nsolid-ai-skills.git) with git over HTTPS, and copies every folder whose name starts with ns- into .agents/skills. Existing ns- skill folders are replaced, so reinstalling updates them.

Skills installation requires git on your PATH and network access to the repository.

Where skills live

After installation, the skills are in the workspace under:

.agents/skills/

Because the skills are local to the workspace, the extension can load them when you start a supported workflow. Other AI agents that read skills from .agents/skills can use them too.

Chat commands that use skills

Type @nsolid in the chat panel followed by a command. These commands load a skill from .agents/skills when it's installed:

Command	Skill folder	What it does
/analyze-asset	ns-analyze-asset	Analyzes an existing N|Solid asset or local profile file.
/analyze-event	ns-analyze-event	Investigates an N|Solid event with runtime context.
/analyze-cpu	ns-analyze-cpu	Investigates CPU-related performance problems.
/analyze-memory	ns-analyze-memory	Investigates memory growth and heap usage.
/benchmark-run	ns-benchmark-run	Benchmarks a single function and saves the result.
/benchmark-validate	ns-benchmark-validate	Compares original and optimized implementations with A/B benchmarks.
/audit	ns-audit-dependencies	Audits workspace dependencies and suggests remediation.
/upgrade <package>	ns-upgrade-package	Evaluates and guides a package upgrade.
/replace <package>	ns-replace-package	Finds a safer or better-maintained replacement package.
/node-upgrade	ns-node-upgrade	Plans a Node.js version upgrade.
/generate-asset	—	Captures a new CPU profile, heap snapshot, or heap sample.

The dependency and Node.js upgrade commands also use the NCM MCP server directly. See NodeSource Certified Modules.

Where these workflows start

You can type these commands yourself, but most workflows start from an extension view that passes context into chat:

• Asset analysis — Ask AI in the Asset Directory.
• Event investigation — Analyze with AI in the Events view.
• CPU and memory analysis — Ask AI on an alert in the Telemetry view.
• Benchmarking — the Generate Benchmark code lens above a function.
• Dependency review — the actions in the NCM Actions view.