00:00:00This is ContinueDev. It's not copilot or clod code and it doesn't replace you, it just automates the parts that are killing your speed.
00:00:07Code reviews are supposed to make code better, but lately they're just slowing everything down.
00:00:11So the obvious answer is to let AI handle it, except that's not actually the problem.
00:00:16The real problem is everything around the code. We have videos coming out all the time, be sure to subscribe.
00:00:27To set up ContinueDev, first install the extension. So I just open VS Code here, I'm going to go to extensions, search continue, install, and reload when I need to.
00:00:35Now the continue icon shows up in the sidebar.
00:00:38Next, we can connect a model to this just by opening settings, edit your config file if it's not already set up, but really you can just add a model in like this.
00:00:47If you want cloud, add a model like GPT4 or clod with your API key, save, and reload. That's it.
00:00:54Or you can even go local here to make all this free using a llama. Pull a model like llama3 and then add it to your provider.
00:01:01And you can switch models anytime from the sidebar dropdown. Cloud, local, whatever you're already using without lock-in.
00:01:09Now let's say you get dropped into a new codebase. You open a project folder and there are things that you don't quite recognize.
00:01:16Nowadays, many of us just copy and paste this, open a browser tab, paste it in a clod or open AI, and then add a bunch of context manually.
00:01:24Then maybe we still have to paste in a few files just so it understands what's going on.
00:01:29And even then, the answers are rather generic. It doesn't really know your project because it's not inside your project.
00:01:36Now we can just ask continue. I can go into my code here and I can highlight something with control L and just type explain this in the sidebar chat.
00:01:45The part you don't expect is it doesn't answer in isolation. It pulls context from your whole codebase.
00:01:51Open files, even git diffs, and gives you an explanation that matches your project as a whole.
00:01:57Sure, honestly, clod code can do this, even cursor. So I feel like this is best if you don't actually rely on AI that much, or if you need a faster tool to cross-check your codebase.
00:02:08Now, here's the important part. When continue suggests a change, like refactor this function, it doesn't just describe it.
00:02:16It generates a real diff preview right in the chat, you review it side by side, then you can apply the changes.
00:02:22You can change it or reject it, but really you stay in control of what's happening, not just cursor taking over.
00:02:28But this is where things change. Normally a pull request just waits. So instead of waiting, you see a review agent.
00:02:36You can create a file like my review agent, just a markdown file, write some rules you care about like formatting, logic issues, security rules, whatever good code means for you.
00:02:47Then you can just run this. So in the chat, I can type agent my agent review and run this PR.
00:02:54Or you can use a pre-built one like continue dev for security. It scans instantly and suggests fixes as comments or diffs.
00:03:02And it's not just in the editor, right? Same agent, no editor. You can manage agents, set triggers, connect slack if you want to for PR alerts, and they land right in your channels.
00:03:13It's not only for coding, it's for shipping faster. Now it's not perfect, but for being a free tool that still lets me do what I want, just giving feedback, it's a really nice extension.
00:03:24And it's free when you use Olama, not OpenAI. So who is this actually for? That's the question here. I guess this is best for a few types of people.
00:03:32If you're working solo, it's just a second set of eyes that never gets tired. If you're on a small team, it keeps standards without turning every code review into some kind of debate.
00:03:41Plus, if you're on VS Code, this fits really well. Now, people aren't using continue because it writes code for them. They're using it because it removes friction. There are fewer review cycles, cleaner pull requests, faster merges.
00:03:55You don't give up control here. The AI just handles the repetitive parts you already hate doing, let's face it. Plus, it's open source. Like I said, it's free when you use a specific model, otherwise you have your API keys, and you can install it and get started in under a minute.
00:04:10At this point, the comparison is obvious. Now, how is this different from Copilot, right? Copilot's in VS Code. GitHub Copilot is great at one thing, helping you write code faster.
00:04:20Continue does something different. Copilot helps you type. Continue helps you ship. Copilot lives in your editor. Continue lives across your whole workflow. Pull requests, CI, agents.
00:04:32And they're not competitors, although they might feel the same on the surface. They work well together. You can use Copilot for autocomplete, and then you can use continue to make sure your code is actually ready to merge.
00:04:43Give it a try if you haven't, or if you have already tried this, what are your thoughts? Are you still using it, or have you dropped it? We'll see you in another video.