Run terminal commands directly on the server.

Install

mkdir -p .claude/skills/terminal && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/17149" && unzip -o skill.zip -d .claude/skills/terminal && rm skill.zip

Installs to .claude/skills/terminal

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Execute shell commands on the server
36 charsno explicit “when” trigger
Intermediate

Key capabilities

  • Execute shell commands on the server
  • Run scripts or one-liners
  • Check system state including OS, processes, and files
  • Build or compile code
  • Install or update packages
  • Execute any CLI tool available on the server

How it works

The skill uses the `builtin_terminal` tool to execute a specified shell command directly on the server, capturing its output.

Inputs & outputs

You give it
A shell command string, optionally with a timeout and working directory
You get back
The standard output and standard error from the executed command, potentially including embedded images

When to use terminal

  • Check server state
  • Run build scripts
  • Execute CLI utilities

About this skill

Terminal Skill

Executes arbitrary shell commands on the server (not on user's machine). Useful for:

  • Running scripts or one-liners
  • Checking system state (OS, processes, files)
  • Building or compiling code
  • Running tests or benchmarks
  • Installing or updating packages
  • Any CLI tool available on the server

Use builtin_terminal tool with fields:

  • command (required): shell command to execute (e.g., ls -la /tmp)
  • timeout (optional): execution timeout in seconds (default 30, max 300)
  • workdir (optional): working directory for the command

When the command generates image files (e.g., output.png, screenshot.jpg), always return everything in a single message. Embed the image inline via ![alt](data:image/png;base64,...) — do not split into multiple messages.

⚠️ WARNING: This tool runs commands directly on the server with the full permissions of the process. Every invocation requires explicit user approval.

When not to use it

  • When the command needs to be executed on the user's local machine
  • When the command requires interactive input not supported by the tool
  • When the command might compromise server security without explicit approval

Limitations

  • Commands run directly on the server with full process permissions
  • Every invocation requires explicit user approval
  • Default timeout is 30 seconds, maximum is 300 seconds

How it compares

This method allows direct server-side command execution, providing capabilities beyond local file operations or code generation.

Compared to similar skills

terminal side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
terminal (this skill)02moNo flagsIntermediate
bazel-build-optimization143moNo flagsAdvanced
github-actions-templates74moNo flagsIntermediate
linux-production-shell-scripts76moReviewIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry