A practical, step-by-step GitHub post for beginners: MCP servers that make AI coding faster, safer, and more reliable. This repo can be divided into two: a simpler version: README(this file), and for the complete installation and usage guide, please refer to
details.md. You can jump to read as needed.
This repo exists to turn AI coding from “autocomplete on steroids” into a tool-using development workflow. MCP is the standard bridge that lets an AI client call external tools and pull real context—so your agent can fetch the latest docs, inspect a running browser, and operate GitHub actions instead of hallucinating.  If you’re new, the goal is simple: copy, paste, run, verify. Coming next: a detailed guide (details.md) with setup checklists, safe key handling, troubleshooting, and prompt recipes you can reuse for any project.
- What MCP (Model Context Protocol) is and why it matters (think “USB-C for AI tools”).
- A curated list of MCP servers (web scraping, search, latest docs, browser debugging, deploy, storage, GitHub ops).
- A curated list of IDE extensions (Claude Code, GitLens, Office Viewer, ESLint, Prettier).
The table below lists the MCP services covered in this post. Each item includes a short description—for step-by-step setup, installation, and runnable examples, please read details.md.
| Plugin / Service | What it does(short) |
|---|---|
| Automatically scrapes and crawls web pages and returns structured content for your AI agent. | |
| Adds real-time, privacy-focused web search capabilities to your AI assistant. | |
| Fetches the latest framework/library documentation to reduce “outdated knowledge” problems. | |
| A Chrome extension: select a UI component on any page and let the AI replicate it using captured DOM/CSS/context. | |
| Lets AI control and inspect a live Chrome browser for reliable automation, debugging, and performance analysis. | |
| One-command deployment to Tencent EdgeOne Pages to get a public URL and speed up global access. | |
| Lets AI operate Tencent Cloud Object Storage (upload/download, manage files, and more depending on configuration). | |
| Official MCP server: AI can read repos and help with workflows like issues/PRs. |
➡️ If you are a beginner or wish to understand the specific usage of each extension, it is strongly recommended to read details.md. This file provides detailed installation steps, configuration examples, frequently asked questions, and sample prompts to help you quickly get started with these amazing extensions. We will also release the second part in the future, introducing more IDE plugins and workflows.
MCP servers can access your tools/data. Only install from trusted sources, use least privilege, and rotate keys if needed. (There have been real-world malicious / vulnerable MCP server incidents.)
See: security notes inside the full post.
- MCP official overview: modelcontextprotocol.io
- Firecrawl MCP, Brave Search MCP, Context7, Web-to-MCP, Chrome DevTools MCP, EdgeOne Pages MCP, Tencent COS MCP, GitHub MCP server
- VS Code extensions: Claude Code, GitLens, Office Viewer, ESLint, Prettier