Build a Gemini CLI check for technical mistakes

A developer made a Gemini CLI command that scans blog drafts for technical errors and runs again after each revision.

Done with Gemini· Intermediate

What they did
Connie Leung asked Gemini CLI to draft a custom command in TOML, then cleaned it up by hand and stored it in .gemini/commands. She set it up to act like a DevRel editor, then ran it on a post draft either by piping the file in or by pointing Gemini at @draft.md with Gemini 3 Pro Preview. After each pass, she edited the draft and ran the command again.
How it went
The check caught three issues in her Angular/Firebase article: a blank screen if startup failed, a dev-mode fetch interval that was too long, and validation that quit after the first missing environment variable. She corrected all three herself. The command is only reviewing the code described in the post, so each suggestion still needs a human check.
Tips
A setup like this works best as a repeatable review step, not as an automatic fixer. Keep the command focused on reporting problems, then verify every suggested change before you commit it.

Try it yourself with Gemini

Create a Gemini CLI custom command named /critique that reviews a blog draft at [path or @file] for technical mistakes, including wrong facts, bad code, outdated guidance, and unclear explanations. Act as a DevRel editor and return a numbered list with the exact passage, why it is a problem, and a suggested correction. Do not change any files. If any fix would require deleting, submitting, sending, or otherwise making a permanent change, stop and ask for my approval first. After I revise the draft, be ready to run the check again.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on dev.to ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.