WEBVTT 00:00.040 --> 00:00.480 Hey there. 00:00.520 --> 00:01.240 Eden here. 00:01.240 --> 00:06.760 And this is a setup video for our boilerplate skeleton for our project. 00:06.760 --> 00:11.440 So we're going to implement now the MCP link chain adapters. 00:11.440 --> 00:13.960 And we're going to learn a bunch on MCP clients. 00:14.280 --> 00:18.080 So in this video we'll create a new project with UV. 00:18.280 --> 00:22.600 We'll create a virtual environment and we'll install our dependencies there. 00:22.920 --> 00:25.560 And then I'm going to commit it to GitHub. 00:25.960 --> 00:32.480 If you want to start from after I'm committing here this code and have the exact code that I have, 00:32.880 --> 00:36.440 then you can simply go and run those commands over here. 00:36.520 --> 00:41.120 So this is going to clone the repository of the course. 00:41.240 --> 00:45.960 It's going to clone the project slash LinkedIn MCP adapters branch. 00:46.280 --> 00:54.280 And after we do that we want to go and CD into the Linkchain MCP adapters where all the code here. 00:54.280 --> 00:58.880 And then we want to check out a commit that I used here. 00:59.120 --> 01:02.680 So this commit has all the code in this video. 01:03.160 --> 01:05.880 And don't worry about copying this command. 01:05.920 --> 01:07.800 You can download the exact command. 01:07.800 --> 01:10.120 I'm going to link it in the video resources. 01:10.400 --> 01:14.680 So this is in case you do not want to start from zero like I'm doing. 01:15.960 --> 01:16.440 All right. 01:16.440 --> 01:24.360 So I'm going to start and go to my GitHub projects and let me clone the repository of the MCP crash 01:24.360 --> 01:24.880 course. 01:24.920 --> 01:28.080 And the URL is going to be listed in the videos resource. 01:28.080 --> 01:29.840 So don't worry about it. 01:30.160 --> 01:30.680 All right. 01:30.680 --> 01:35.120 So I cloned the repository and let's CD into this repo. 01:35.960 --> 01:38.200 And currently I'm on the main branch. 01:38.200 --> 01:39.480 And let me show you it. 01:41.840 --> 01:47.160 And we want to create now a new branch which is going to be for this project. 01:47.280 --> 01:56.320 So let me go write git checkout and I want to write dash dash orphan to make it detached from my original 01:56.320 --> 01:56.840 repo. 01:57.040 --> 02:01.560 And let's call it project slash link chain MCP adapters. 02:02.560 --> 02:06.000 We created the branch and we're on the new branch right now. 02:06.040 --> 02:10.320 Let me show you now the current working directory of the project. 02:10.960 --> 02:15.080 And now it's time to see what files do we have here. 02:15.680 --> 02:17.680 So we see we have a bunch of files. 02:17.680 --> 02:21.360 And you're probably going to see the repo where it's much more organized. 02:21.360 --> 02:23.120 So don't worry about the files here. 02:23.360 --> 02:27.960 What we want to do is to remove those files and start from a clean slate. 02:28.160 --> 02:30.080 So let me go and do that. 02:31.080 --> 02:35.400 So let me write here git rm rf dot. 02:35.400 --> 02:38.760 So this is going to remove all of those files not to track them. 02:38.760 --> 02:40.840 And right now we don't have anything. 02:40.840 --> 02:44.120 And I listed all the files and I don't see anything here. 02:44.120 --> 02:45.720 So this is our starting point. 02:45.720 --> 02:48.920 We have a new branch and we don't have any files here. 02:49.120 --> 02:49.880 Alrighty. 02:49.880 --> 02:53.560 So let's go and initialize our project with UV. 02:53.680 --> 02:59.360 So I wrote uv init, and we have now a bunch of boilerplate files that we can work with. 02:59.560 --> 03:07.200 We can see that the files that were created is a empty readme main.py and a pyproject.toml file. 03:07.400 --> 03:07.960 Cool. 03:07.960 --> 03:14.800 So now I want to open the cursor IDE and this is now our working station. 03:14.800 --> 03:17.280 So let's go and open up terminal. 03:17.600 --> 03:20.200 Let's now create a virtual environment. 03:20.200 --> 03:22.040 I'll write UV Venv. 03:22.960 --> 03:25.160 And now we have a new virtual environment. 03:25.160 --> 03:27.600 You can see we have a venv here. 03:28.320 --> 03:31.080 And let me now source it and activate it. 03:31.080 --> 03:33.120 So I'm going to copy here the script. 03:33.280 --> 03:37.480 And now you can see on the left side we have MCP Crash Course inside parentheses. 03:37.480 --> 03:39.320 So this is our virtual environment. 03:39.520 --> 03:47.800 And by the way if I'm going to exit cursor and open it again then it should automatically activate the 03:47.800 --> 03:48.760 virtual environment. 03:48.760 --> 03:50.640 You can see here on the left side. 03:50.760 --> 03:53.880 And if not you can simply activate it like before. 03:54.400 --> 03:57.760 Anyways, let's go now and let's install our dependencies. 03:58.080 --> 04:01.800 And if I will go to the link Chain adapters repository. 04:02.280 --> 04:06.120 And we can see here in their example we need a bunch of packages. 04:06.480 --> 04:11.880 We need the link chain MCP adapters package lang graph package and link chain OpenAI. 04:12.240 --> 04:18.240 And notice we are not installing explicitly the MCP package because when we install the Lang chain MCP 04:18.280 --> 04:20.800 adapters package, it installs it automatically. 04:21.080 --> 04:21.800 Alrighty. 04:21.840 --> 04:26.920 So let's go and let's copy now those packages and let's add them with UV. 04:30.640 --> 04:36.600 And after we install them let me also add python dot env package to load up environment variables. 04:36.840 --> 04:40.760 And and if it's not already installed as part of their dependencies. 04:41.120 --> 04:46.200 So let me go and add python dotenv and we can see it's already installed. 04:46.640 --> 04:48.480 And let's see the toml file. 04:48.480 --> 04:51.800 And we can see now those are the dependencies that our project needs. 04:53.200 --> 05:00.960 And if we go and head up to UV dot log file here we have the exact version of all the packages in our 05:00.960 --> 05:03.480 virtual environment that we need for the project. 05:03.480 --> 05:05.920 And for example, here I installed LinkedIn core. 05:05.960 --> 05:08.920 Right now the current version is 0.3.5. 05:08.960 --> 05:10.240 One of LinkedIn core. 05:10.560 --> 05:15.120 And when you install it you might have different versions, probably newer ones. 05:15.280 --> 05:18.320 And I'm going to keep this repository and videos up to date. 05:18.320 --> 05:23.320 So unless there are breaking changes I'm going to leave it that if they are I will update this video. 05:23.720 --> 05:28.320 Alright, so let's go to our main.py file and let's go and try to run it. 05:28.600 --> 05:34.640 Let's see, just as a sanity check that everything is working and I'll run it with you've run Main.py 05:34.960 --> 05:37.440 and notice we're inside the virtual environment. 05:37.440 --> 05:39.800 So it's going to contain all of our packages. 05:40.560 --> 05:41.200 All right. 05:41.200 --> 05:42.600 So let's go and run it. 05:42.600 --> 05:44.880 And boom we get the output that we want. 05:45.080 --> 05:46.520 What a beautiful program. 05:46.800 --> 05:47.360 Alrighty. 05:47.360 --> 05:50.040 So I want to make it async. 05:50.240 --> 05:54.840 So I turn this function into a coroutine with the async keyword. 05:55.080 --> 05:59.320 And now in order to run our main file I need to run it with async IO run. 05:59.560 --> 06:01.910 And I love these cursor autocomplete. 06:01.910 --> 06:04.910 I'm simply pressing tab tab tab and it completes me. 06:05.310 --> 06:06.990 And I forgot to import async IO. 06:06.990 --> 06:07.910 So let's do that. 06:08.190 --> 06:10.950 And let's go now and run the code again. 06:12.150 --> 06:13.710 And we can see it's still working. 06:13.710 --> 06:15.350 And this is a good sign. 06:15.630 --> 06:20.990 And I want to create now a dot env file to hold our environment variables. 06:20.990 --> 06:23.750 Because we're going to be using an LM. 06:23.750 --> 06:26.030 So I'm going to use a managed LM. 06:26.430 --> 06:29.870 And let me paste here the values of my environment variables. 06:30.110 --> 06:34.670 And by the way don't worry about the values here of me exposing them. 06:34.710 --> 06:38.630 Um I'm going to revoke it right after I finish exporting this video. 06:39.070 --> 06:44.910 Anyways, I'm using OpenAI right now, but you can use any LM you want that supports function calling. 06:44.910 --> 06:46.550 So you can use anthropic sonnet. 06:46.550 --> 06:47.990 You can use Gemini. 06:48.470 --> 06:50.390 By the way Gemini has a free tier. 06:50.390 --> 06:52.110 So you can use it as well. 06:52.510 --> 06:56.430 And you can use deep sea or any other LM which supports function calling. 06:56.910 --> 07:02.910 So the first environment variable is going to be our OpenAI API key or our API key? 07:03.510 --> 07:07.990 Now the other four API keys you can see here of linked chain. 07:07.990 --> 07:10.510 They are for the link Smith tracing. 07:10.510 --> 07:15.430 So we can actually trace the LLM calls and see exactly what's being sent to the LLM. 07:15.590 --> 07:16.670 And what do we get back. 07:16.670 --> 07:20.070 So I think it's very important to understand it's not a must for you. 07:20.350 --> 07:27.350 So if you do not trace it you simply need to put the link chain tracing V2 equals to false here and 07:27.710 --> 07:28.630 it should work. 07:28.630 --> 07:32.750 And if you are using simply go to the link chain platform and generate the API key. 07:32.750 --> 07:33.390 From there. 07:33.630 --> 07:38.550 Let me know if you wanted to make a video on that, but I'm assuming that you know how to do it anyways. 07:38.550 --> 07:44.470 So I put my API key, I put the endpoint and I call the link chain project MCP test here. 07:44.830 --> 07:52.590 So even though I did revoke now the API keys so you can't actually use them, I want to add here another 07:52.590 --> 07:53.350 file, let's call it. 07:53.350 --> 07:54.150 Gitignore. 07:54.310 --> 07:58.350 And I want to not commit the file to the repository. 07:58.630 --> 08:04.070 So let me just write here And this is simply going to ignore the dot env file. 08:04.070 --> 08:05.990 And you can see it's also grayed out here. 08:06.550 --> 08:08.590 So git is not going to track this file. 08:08.590 --> 08:12.070 And I'm not going to accidentally upload it to my repository. 08:12.670 --> 08:13.110 All right. 08:13.110 --> 08:15.390 So let's go back to the main.py file. 08:15.390 --> 08:20.910 And now we want to load those environment variables and to see that everything is working correctly. 08:20.910 --> 08:22.070 So we can continue. 08:22.070 --> 08:25.950 And to write our actual code and write our first MCP client. 08:26.310 --> 08:29.990 So I'm going to import from env the load inv function. 08:29.990 --> 08:32.630 And I'm going to call it right before my code runs. 08:33.270 --> 08:39.430 And as a sanity check let's go and print the OpenAI API key just to see that we actually loaded it into 08:39.430 --> 08:40.710 our environment variable. 08:42.030 --> 08:44.430 And let's go and try to run it. 08:44.430 --> 08:47.870 And we can see we forgot to import the OS module. 08:47.870 --> 08:49.590 So let's do that and rerun it. 08:49.590 --> 08:55.790 And boom we can see now our API key and our environment is indeed loaded with the correct values. 08:56.350 --> 08:58.390 So that's pretty much it for this video. 08:58.430 --> 08:59.990 We have all the boilerplate code. 08:59.990 --> 09:03.310 And from now on we can go and we can implement our MCP client. 09:03.670 --> 09:08.790 So let me simply go and add everything and let me go and commit it into our repository. 09:10.230 --> 09:13.230 So now we are tracking all the files we created. 09:13.430 --> 09:19.590 And let me show you one of my favorite cursor features where you simply go and generate automatically 09:19.590 --> 09:21.670 with AI the commit message. 09:22.310 --> 09:24.790 And let me go and commit this code here. 09:30.150 --> 09:33.950 And now let me go and push it into the repository. 09:35.030 --> 09:35.270 Oops. 09:35.310 --> 09:37.670 I need to set my upstream repo here. 09:38.390 --> 09:41.510 And now we should see the code available on GitHub. 09:41.510 --> 09:42.750 So let's go to GitHub. 09:42.910 --> 09:48.990 And if I'm going to select now the branch of project slash MCP adapters. 09:49.310 --> 09:50.990 And let's go and click that. 09:51.710 --> 09:55.270 We can see now all the code that we just wrote. 09:55.550 --> 10:00.510 We can see the main.py file, the UV log, the pyproject.toml file and everything.