You found the script. But what does it actually do?
Shell scripts rarely explain themselves. Makefile targets are cryptic. Even npm scripts chain together enough flags and pipes that you have to read the source to know what happens when you hit run.
CommandTree 0.5.0 fixes that. Hover over any command and a tooltip tells you exactly what it does, in plain language.
How It Works
When GitHub Copilot is installed, CommandTree reads the content of every discovered command and asks Copilot for a one-to-two sentence summary. These summaries appear instantly when you hover:
Compiles the TypeScript extension, packages it as a .vsix file, and installs it into VS Code in one step.
No reading source code. No guessing. Just hover and know.
Security Warnings
Copilot also flags dangerous operations. If a script runs rm -rf, force-pushes to a remote, or handles credentials, the tooltip includes a security warning and the command label shows a warning indicator. You know the risk before you run.
Stored Locally, Updated Automatically
Summaries are cached in a local SQLite database at .commandtree/commandtree.sqlite3 in your workspace. They persist across sessions and only regenerate when the underlying script content changes, so there is no repeated API overhead.
Works Without Copilot
Every core feature of CommandTree, including discovery, execution, tagging, and filtering, works without Copilot. AI summaries are a bonus layer. If Copilot is unavailable, the extension behaves exactly as before.
Get Started
Update to CommandTree 0.5.0 from the VS Code Marketplace, make sure GitHub Copilot is installed, and hover over any command in the tree. For full details, see the AI Summaries documentation.