Gemini CLI in 6 Minutes: Google's Free and Open-Source Coding Assistant

--- type: transcript date: 2025-06-25 youtube_id: T76NbeTdDFA --- # Transcript: Gemini CLI in 6 Minutes: Google's Free and Open-Source Coding Assistant Google has just open sourced their Claude Code competitor, Gemini CLI. In this video, I'm going to be taking a look at what they just released as well as show you how you can get started with it. First up, the big news with this. This is completely for free. You can access in full the million tokens of contacts that Gemini 2.5 Pro has, as well as have up to 60 model requests per minute and up to a,000 requests per day at no charge. All that you need to do to get started is simply log in with your personal Google account to be able to access all of that. Now, if you are a professional developer or if you just want a higher rate limit than what they give you, you do have a couple options. You can get your Google API key from the Google AI Studio or from Vert.Ex AI. Alternatively, you can get a Gemini code assist standard or enterprise license where it is built at a monthly rate to be able to access a much higher rate limit. Now in terms of some of the specifics and tools that come equipped with the CLI does ground prompts with Google search. So you are going to be able to fetch information from web pages as well as provide real-time external context to the model. This can definitely be helpful in a coding context especially with libraries changing and versions of different things constantly changing. Being able to have that ability to search the web is definitely very handy. So it does also come with built-in support for MCP or model context protocols. So you can add in any MCP from the growing ecosystem there. They also mentioned that you can automate tasks and integrate with existing workflows by invoking the Gemini CLI non-interactively within your scripts. This is potentially interesting. What comes to mind with this is Claude recently came out with their SDK within both Typescript as well as Python. I wonder with something like this if that's what they're leaning into. Also within the blog post, they have a demonstration of what looks to be an MCP server where it's generating both videos as well as images from Google's flagship both image as well as VO models. In terms of the license, the Gemini CLI is fully open source with an Apache 2 license. You're welcome to inspect and understand exactly how the code works. And that is a notable difference from Claude Code where that is closed source. Now, similar to Claude Code having the Claude MD, there is now a Gemini MD. It doesn't look like there is quite a standard yet in terms of the naming convention for actually passing in effectively a system prompt into these CLI tools. That certainly would be nice if there is a standard like an agent MD or something like that that can be ported between Gemini or Cloud Code or what have you. But nonetheless, it is nice that it does have that capability to look for that context if you have further instructions within your Gemini markdown file within your repository. One quick aside with this is they do have Gemini code assist. I haven't personally used this before, but by the looks of it, it does look very similar to something like Cursor Agent. If anyone has used this, I'd be curious your thoughts within the comments below how it performs to something like Cursor or Windsurf or GitHub Copilot. It does look like it does have that agentic mode where it can go through and do the multifile edits and all of that, but I haven't personally used this quite yet. Now, in terms of getting started, I'll link this within the description of the video. And these steps are super simple. All that we need to do is make sure that we have at least Node.js JS version 18 installed. You can run this with this MPX command where it references the GitHub repository. What I'd encourage you to do is to install it globally with the CLI command here. npm install Google CLI. What will happen as soon as you install it? You'll be able to log in with your personal Gmail account. Once you've logged in, you can set the theme and then at any point in any project, you can run Gemini and you'll be able to interact with this. So what I'm going to do in the root of this project that I just initialized it in is I'm going to create a new Nex.js application and I'm going to say I want to update the homepage to read developers digest. One quick note with this is it works similar to something like cloud code where if you click shift tab you can accept the edits within here. If it does need to write new files or edit existing files by enabling that shift tab it will go ahead and not prompt you with each file that it's asking to edit. So within here it's starting to understand how our project is structured. So it's reading through the directory here. I can see that it read the page.tsx as it exists now within the next.js template. And from there it immediately went through and it updated our homepage. Now what I'm going to do here is I'm just going to start our development server. So we can take a look at what it's doing here. And then here we go. Then from here you can begin to use it as if you were using something like cloud code or cursor agents. in this case, just to quickly test it one more time. I'll say I want to build out a beautiful SAS landing page that replaces the homepage. One quick note with this is Gemini 2.5 Pro is quite fast. I've been using Claude for Opus over the past several weeks and while it does give quite good results, it definitely is quite a bit slower than Gemini 2.5 Pro. One notable difference with Gemini is in terms of the UX, they've definitely gone with a more verbose default for what they actually show you in the responses within here. I can actually see what the agent is doing throughout all of the different files. I can also see the thinking process between everything that it's doing here. With this, if it is editing pre-existing files, you will be able to see the diff of what it's removing as well as what it's adding within here. I'd imagine that you would be able to configure certain things to auto accept like npm commands and all of that as well. And now, if I restart my development server, here is what it generated for us. We have this SAS landing page. in just really a number of seconds is what it generated for us. And mind you, this is completely for free and open source. That's pretty much it for this video. I just wanted to give you a really quick look at the Gemini CLI. Kudos to the team at Google for both making this free as well as open source. That's it for this one. If you found this video useful, please comment, share, and subscribe. Otherwise, until the next one.