What it does ai-cli is a thin oclif wrapper around OpenAI’s API. You type ai ask "list processes on port 3000" and it prints a shell command back. It stores your API key locally, supports model switching, and has shell autocomplete.

The interesting bit The author did the boring plumbing so you don’t have to: key management via ai auth, model selection via ai model, and autocomplete setup via ai autocomplete. It’s essentially a well-packaged prompt with a system message, but the packaging is what makes it usable.

Key highlights

• Built on oclif, so it gets --help, autocompletion, and update channels for free
• Defaults to gpt-3.5-turbo; you can swap models without touching config files
• Claims ~$0.0009 per command at current token counts
• Hacker News #3 with 159 points, so the itch is real

Caveats

• You bring your own OpenAI API key and billing
• The README still says “GPT3” everywhere while defaulting to 3.5-turbo; branding drift is visible
• No evidence of local caching or offline mode — every command hits the API

Verdict Grab it if you constantly context-switch from terminal to browser for shell syntax. Skip it if you already have a shell alias that curls OpenAI, or if you work offline.