Practical guides · 01
Three checks before you run an AI-generated command.
Check where it will run, what it will touch and what result to expect.
3 minute read
An AI-generated command is a suggestion worth reading. Before pressing Return, take a moment to connect the words you asked for with the command you are about to run.
1. Check the starting folder.
A command that uses . starts from the terminal’s current folder. That is useful when you are in the project you intended to inspect. Use pwd to see the current path and compare it with the folder you meant.
For a first exercise, choose a small project containing no private or sensitive material. Ask for a command that lists its Python files without changing anything.
2. Read the action.
One possible command is:
find . -type f -name '*.py' -printIt searches below the current folder, matches regular files whose names end in .py, and prints their paths. The quotes around *.py leave that pattern for find to interpret. This example lists files; it does not delete or rewrite them.
The generated suggestion may be different. Read that actual suggestion. If it includes deletion, a network request, a permissions change or something else you did not request, stop and revise it before execution.
3. Decide what a useful result looks like.
Before running it, decide what result would satisfy the request. For this example, expect only paths to Python files in the intended project. After the command finishes, compare the actual output with that expectation. An empty result might be correct; it might also mean the starting folder or pattern was wrong.
Try a small task in AITerm.
In AITerm, plain-English requests lead to editable command proposals. Review the proposed command before running it, use /explain to understand output, and use /fix after a command fails to get a suggested correction.
The core app is free on Apple Silicon Macs with macOS 13 or later. It includes the command loop, tabs, split panes, saved workspaces, Ollama and supported bring-your-own-key connections. An on-device model needs an initial setup download; cloud connections use your chosen provider and may have separate usage charges. Pro adds longer workflows such as agents and runbooks.
Keep learning.
- What is included in Free and Pro
- How local and cloud AI handle your data
- Safety settings and command review
Prepared with AI assistance from the current AITerm documentation. This command is an illustrative example, not a recorded app demonstration.