sandbox-terminal

What problem does it solve?

This Skill guides AI agents to safely, efficiently, and correctly use terminal commands in any workspace. It ensures actions stay within the environment, avoids secrets leakage, and defers to workspace-specific steering docs for concrete commands.

Core Features & Use Cases

• Safe, token-efficient terminal usage: agents learn to run commands without disrupting the workspace and with minimal output.
• Environment detection and runner resolution: detects the project root and selects the correct runner (e.g., uv run, poetry run, pnpm, etc.) before execution.
• Workspace overrides and portability: consults {{workspace}}/.github/ for organization-specific rules and scripts, enabling portable automation across workspaces.

Quick Start

Activate when the user asks to run a command, install dependencies, run tests, start servers, lint, build, create scripts, debug terminal errors, or compose CI/CD pipelines. The agent should first perform environment detection, then check for workspace steering docs, and finally execute the user's request using the appropriate runner.