コンテンツにスキップ

MCP Tutorial: Integrate Atlassian, Jira, and Confluence

  • URL: https://www.youtube.com/watch?v=pXAih3jAOcc
  • 文字起こし日: 2026-02-24 02:40

内容概要

  • MCP（Model Control Protocol）は、LLM（大規模言語モデル）にコンテキストを提供する統一された方法を提供する。
  • MCPクライアントは、カレンダー、メール、ローカルデータストレージなどの複数のサーバーに接続できる。
  • 通常のAPIとは異なり、MCPは複数のAPIを設定する複雑さを軽減し、単一APIでコンテキストを提供する。
  • MCPの機能には、シングルプロトコル、動的なツールディスカバリー、双方向通信、リアルタイム通信、スケーラビリティ、セキュリティ制御が含まれる。
  • MCPサーバーはクライアントが利用でき、ホストにコンテキストを提供してアクションを実行する。
  • Atlassian製品（Jira、Confluence）との連携には、MCPサーバーの公式リポジトリを使用する。
  • AtlassianアカウントのAPIトークンを取得し、MCPサーバーを設定する。
  • UVという新しいパッケージマネージャーを使用してMCPのAtlassianモジュールをインストールする。
  • VScodeやCursorなどのIDEでMCPクライアントを設定し、Jiraのタスク作成や削除などのアクションを実行する。
  • MCPを活用することで、LLMがコンテキストを理解し、よりインテリジェントな応答を生成できるようになる。

文字起こし

Hello all, welcome to this video. In this video we'll talk about MCP and how can you integrate Atlassian's MCP server and use it like for Jira and confluence. With a simple prop you can execute actions on a cloud. So let's see how it goes. So first of all, let's talk about MCP. MCP is a model control protocol which provides an unified way to provide the context to LLM. Like as you can see in this image, there are multiple MCP servers for there is a separate server for calendar, mail, Slack and local data storage. We connect that all these servers to the MCP client and that with the help of MCP host executes action. Uh the difference between a normal API and using an MCP is that uh without an MCP, we need to configure multiple APIs and it becomes very complex to provide the context with a single API. That's why we use MCP. Here you can you can see the differences between MCP and API. like there is need to separately configure all the APIs. uh and MCP provides a single uh way, a single and standardized way to connect, then real time communication is possible in MCP, dynamic tool discovery, scalability and security and control that are the differences in MCP and traditional APIs as you can see. Let's talk about features of MCP, single protocol, dynamic tool discovery and two way communications. Uh here is the architecture of MCP as you can see. Uh let's first talk about the local data source file. Uh local data source files and remote service uh we can access it by a simple API and it all are integrated into one MCP server. That MCP server can be used by clients and clients which provides the context to the host and it executes the excel. here are some uh resources that I'll put in the description you can see. Now we'll talk about how you can integrate uh uh MCP servers official repository. Here they have the various servers which are official servers some community server. In that there is one server for the Atlassian. We'll use this to connect conference and Jira. So this is the repository. Uh as you can see in this video, you are just prompting the action that you want to take and it automatically executes such thing like creating task updating task etc. So for that you will need an API token of your Atlassian account. Okay, so just go to this uh this URL uh you will get an API tokens from here. Just create it. and uh you need to then once you created the API token, you need MCP servers official repository. Here they have the various servers which are official servers and community server. In that there is one server for the Atlassian. We use this to connect conference and Jira. So this is the repository. Uh as you can see in this video, you are just prompting the action that you want to take and it automatically executes like creating task, updating task etc. So for that you need an API token of your account. Okay, so just go to this uh this URL you will get an API tokens from here. Just create it. And you need to then once you created the API token, you need to install this library with the help of UV. So what is UV? UV is a new package manager uh which is built in rust and provide a blazing speed for Python package management. Once you have installed this MCP Atlassian you atlassian's module then just configure this if you are using a configure this by if you are using a cloud just your company's name and Atlassian URL your personal email and the API token that you have created by visiting that link. So that one and for server and data centers you just know the API token. There are two types of transport in MCP. Uh one is SSH transport and second one is STDI transport. So SSH transport is useful when you want to connect you want to connect with a cloud or a remote server. and STDI when you have a local setup and you want to connect there. So uh you can choose the options uh which you want to choose. Uh you can use uh SSC and uh STDI as well. Once you are done this, just copy this command. And now we will use VS code. Now we'll see now we'll see how can you integrate that server in VS code. Just go to an extension and search for client uh and install it. Once you have installed this extension, you can see a there is an client icon in VS code's right side. Just open it and here once you restart, you'll see the two options. The one is to get login and use free tribute or just enter your API key and visit this. So I have entered my open API here are multiple API providers that you can choose and you can see O for the usage of local and various are available now to integrate MCP server of just come to here. Here you will see the MCP. It provides the various MCP servers uh client and server. When I search for Google, there's a server for Google calendar and all. uh if I search for notion there's a MCP server for notion, you just install it and client will handle all the back end. you don't need to configure. Uh as we have installed it in our directory. we will add our MCP server. just as I told you, there are two types of transfer. one is remote transfer when you are using a cloud or and cloud or a remote uh desktop and there is one second option local command then uh in when in it that you have that MCP server is installed in your local desktop. Just so just click on this client settings.json and paste the thing that we have copied from the GitHub repo of that and change the just change your API key and all. And once we have created this, just close it, save it and close it. Okay? Then just refer GitHub. Now, here is type something that you want to see. Uh like I see the create an issue in Jira for a project with project name is Jira MCP test and with description of is for Jira's MCP server. This is you can see the request is there and it will ask us to approve the request. Yes we this is there and identified all the information that we have provided in the pro. just approve it and now it will provide the create the yes it has created an issue in GitHub. Let's see if it has been created or not. Oh so this is my Jira. this yes you can see there it has created the Jira test. It has been just created and you can also delete this uh issue. So what's the key for this 8020. I say uh delete issue with key 1820 from project with project team 80. Now with this it can it will delete this uh issue that we have created. Yes, has identified that we want to delete that and we just need to approve here. As once we approve this, yes, the task is completed and just refresh this place to see the issue is deleted on there as the issue is successfully deleted. There are multiple actions that you can get with with Jira like getting the issue of specific Jira issue, uh searching for issue, getting project issue, getting epic issue. There are multiple uh Jira actions that you can take from this uh just by a from. So it and if you want some more detail, just provide more details that it can use. Now we'll see uh how you can be implemented in cursor. So cursor is an AI ID. Uh it provides an inbuild search function or a chat that will help us to code. Now to add a MCP server, just go to cursor setting, come to MCP and click here add new global MCP server. It will get us to MCP.json file. There the same thing that we have pasted in lines MCP JSON file. Uh just change the API token and all. And uh and uh it it can show that the server is available. We can try get we can leave it from like get information of of to issue project project you can also add uh your project name also work with that. Now it has identified the what action to take just click on the run tool and it will use the top issue. to add some more functionality to this server just fork this repo, add your code if it's for Jira or uh just I'll show you just go to Jira add here your code what is in issue project and all and just uh once you added your code and the same thing. Just go to that directory, initialize this server and you will be able to get all your custom actions performed will be MCP solved.