Today, I want to share a brief list of MCP servers that I use daily in my work and projects.
1. GitHub MCP Server
What It Does
GitHub MCP Server enables you to work with GitHub through an AI assistant: view code, search for files, read and summarize documentation in repositories, manage issues and pull requests, and get an overall view of projects.
How I Use It
- I read and create documentation in repositories.
- I add to or create new technical documentation.
- I analyze pull requests (what has changed, what risks are involved, what needs to be amended) with the help of rules that impose code style restrictions.
- I create pull requests.
- I look at issues that are relevant to me to understand what to prioritize. This significantly reduces the number of switches between the IDE and the browser, especially when you need to quickly dive into several PRs at once.
Who it's suitable for
- Developers who "live" in the IDE and dislike constantly opening GitHub in the browser.
- Those who work extensively with PRs and technical documentation.
Links
Atlassian MCP Server (Jira / Confluence)
What it does
The Atlassian MCP Server provides access to Jira and Confluence through an AI assistant. You can create and update tasks, read pages in Confluence, search for information, and receive summaries of documentation and backlogs.
How I Use It
- I read large Confluence pages and receive concise summaries instead of manually scrolling.
- I write and refine technical documentation, such as design documents, meeting notes, and instructions.
- I create tasks in Jira (bug, task, sub-task) from simple text notes.
- I edit existing tasks, including descriptions, acceptance criteria, status, and priorities.
- I quickly review tasks that are specifically relevant to me, such as those in progress, those "to do," and those that are blocked. In fact, this is my primary way of working with Jira without delving into its cumbersome web interface.
Who it's suitable for:
- PMs, BAs, Team/Tech Leads who "live" in Jira and Confluence.
- Engineers who want to see "what is expected of me" in a convenient text format.
- Teams with a lot of documentation and lengthy Confluence pages.
Links:
Figma MCP Server
What it does
Figma MCP Server integrates design context into the assistant, including frames, components, styles, tokens, and layouts. It helps transfer designs from Figma to code by understanding the screen structure, spacing, component names, and connections between elements.
How I use it
- It significantly simplifies the writing of components, especially when the screen is complex.
- I obtain the frame structure and can easily transfer it into code.
- I make fewer mistakes in details: consistent spacing, correct colors, and text styles. As a result, the time from "design is in Figma" to "screen is functional in code" is reduced, and the implementation quality is closer to the original design.
Who it's for
- Frontend developers who work extensively with Figma.
- Designers and engineers who want to minimize manual adjustments when transferring designs to code.
- Teams with a complete design system and tokens.
Links
Goman.live – MCP Localization Manager
What it does
Through MCP, Goman.live assists with localization: creating new keys, viewing existing ones, obtaining draft translations in multiple languages without manually editing JSON files, and translating texts without losing context.
How I Use It
- During vibe coding, when it's clear that the text will need translation.
- In projects that require localization into multiple languages (I have several such projects).
- When new keys need to be quickly added and a draft translation obtained, which can then be properly reviewed on the service page.
At the moment, I am satisfied: the localization process has become more predictable and less stressful. However, it is important to monitor the relevance of the keys, as agents tend to create a lot of irrelevant ones.
Who it's suitable for
- Teams that require internationalization (i18n) and support multiple interface languages.
- Developers who prefer not to manually maintain JSON translation files.
- Localizers and content teams seeking a centralized location for translations.
Links
Playwright MCP
What it does
Playwright MCP allows you to control the browser through an assistant based on Playwright: open pages, fill out forms, click buttons, check tables, and read texts on the page. It works with the semantic structure (a11y tree) rather than "screen images," making it more stable in its operations.
How I Use It / When It's Useful
- For simple smoke tests of main user scenarios (login, checkout, order form, etc.).
- For quickly checking that key flows haven't broken after changes.
- For gathering information from pages without writing full e2e tests manually.
This isn't a magical replacement for all tests, but it's a good tool to give agents access to the browser and automate routine checks.
Who It's Suitable For
- QA professionals who are already using Playwright.
- For developers who want to easily verify main scenarios after changes.
- For analysts who need to automatically extract data from web pages.
Links
Brief Conclusion
- Atlassian MCP — My main tool for documentation and tasks (Jira/Confluence).
- GitHub MCP — For code, PRs, and repository reviews.
- Figma MCP — For implementing designs and staying close to the layout.
- Goman MCP — when a project requires localization in multiple languages.
- Playwright MCP — used for UI automation, smoke testing, and verifying main user scenarios.
Top comments (0)