MCP Server to Monitor CI (Continuous Integration) builds

This project aims to demonstrate (as a POC) how we can use MCP Servers to feed real time CI Build context to the IDE and use AI Agents for on the fly status monitoring and context based quick code fixes. The aim of such tool is to enhance the developer experience and cut the time involved in juggling between different tabs

Requirements

1. Docker

2. Git

3. Ngrok

(If Docker Desktop or setup guide has issues running on Windows systems, its recommended to use the setup in wsl)

Steps to Setup Locally

1. Fork the repository

2. Clone the forked repository

3. In project directory run command: make check_deps

4. If any dependency is not installed fulfill the requirement

5. Then run command: make ci-up

6. This command will get the Jenkins running on localhost:8080

7. Run command docker logs contrihub-jenkins

8. Copy the admin password to clipboard

9. Open localhost:8080 on browser

10. Enter the password

11. Choose Install Suggested Plugins

12. Back in terminal run following command on another instance of the terminal: make ip-tunnel

13. This will provide you the public IP endpoint of the localhost:8080

14. In the forked repository on github, go to repository settings and then webhooks

15. Create a webhook and enter this public_IP_endpoint/github-webhook/ as webhook url

16. Open the jenkins portal on browser and under the manage jenkins (settings icon) click on plugins

17. Install following plugins by searching in available plugins: a. github integration b. docker pipeline

18. Configure a new job by clicking on jenkins icon and selecting add new job

19. In configuration page, under triggers select Github hook trigger for GITScm Polling

20. Under Pipeline select the dropdown option as Pipeline script from SCM

21. Under SCM Dropdown select git

22. Add repository URL

23. Select add credentials and add your github username and in password add your PAT token (generated on github.com)

24. Select the added credential by dropdown showing none

25. Make a code change and push your code to github, the build should trigger

Objective

Create the MCP Server with logic to fetch build status and incase of failure fetch the build logs and provide as context to LLM At the end of project, the build must run and succeed

Resources

1. MCP - https://modelcontextprotocol.io/docs/getting-started/intro

2. Jenkins - https://www.jenkins.io/

3. WSL - https://learn.microsoft.com/en-us/windows/wsl/install

4. Docker - https://www.docker.com/

5. Ngrok - https://ngrok.com/

6. CI - https://aws.amazon.com/what-is/ci-cd/

7. Cloud Native Mindset - https://dev.to/aws-builders/embracing-cloud-native-mindset-1fi8