Best AI tools for developers

Most developers already have a coding assistant. What separates a useful one from a frustrating one is what it can reach: current documentation instead of a stale memory of it, the whole codebase instead of one open file, the issue tracker, the browser where the bug shows up. This page lists the MCP servers and agent skills that give an assistant that reach.

Extend the agent you have before you switch

Switching from one coding agent to another costs a week of new habits. Adding a documentation server or a code search server to the agent you already use costs one command, and it fixes the most common failure: the model writing code against an API that changed last year.

For the agents themselves, see the AI coding tools page. This page is about what you plug into them.

How to judge a developer MCP server

Check who publishes it. An official server from the company behind the service, such as GitHub's own, tends to keep up with the API. A community server can be better, but look at the last commit date and the open issues.

Check what it can touch. A server that can run shell commands or write files can do real damage on a bad prompt, so prefer servers with read-only modes, and scope tokens to the repos they need.

Count your servers. Every server adds tool descriptions to the model's context, and a dozen idle ones make every answer slower and vaguer. Keep the three or four you use each day.

The list

From SkillScout's catalogue, counted on 27 October 2026: 271 entries matched this topic, and these are the 12 with the most GitHub stars, one per project, after a review for false matches. Descriptions are each project's own. Stars show popularity, not fit.

  1. context7

    Open-source tool · GitHub · 54,865 stars

    Context7 Platform -- Up-to-date code documentation for LLMs and AI code editors

    Get the code, then follow its README: git clone https://github.com/upstash/context7

  2. github-mcp-server

    MCP server · GitHub · 29,602 stars

    GitHub's official MCP Server

    Add it to Claude Code (from the official MCP Registry): claude mcp add github-mcp-server -- docker run -i --rm ghcr.io/github/github-mcp-server:2.0.3

  3. repomix

    Open-source tool · GitHub · 24,448 stars

    Repomix is a powerful tool that packs your entire repository into a single, AI-friendly file. Perfect for when you need to feed your codebase to Large Language Models...

    Get the code, then follow its README: git clone https://github.com/yamadashy/repomix

  4. code-review-graph

    Open-source tool · GitHub · 15,637 stars

    Local knowledge graph for Claude Code. Builds a persistent map of your codebase so Claude reads only what matters, 6.8× fewer tokens on reviews and up to 49× on daily...

    Get the code, then follow its README: git clone https://github.com/tirth8205/code-review-graph

  5. claude-context

    MCP server · GitHub · 12,127 stars

    Code search MCP for Claude Code. Make entire codebase the context for any coding agent.

    Get the code, then follow its README: git clone https://github.com/zilliztech/claude-context

  6. inspector

    MCP server · GitHub · 10,174 stars

    Visual testing tool for MCP servers

    Get the code, then follow its README: git clone https://github.com/modelcontextprotocol/inspector

  7. open-code-review

    Agent skill · GitHub · 9,619 stars

    Open-source & free, Battle-tested at Alibaba's scale. Hybrid architecture code review tool: deterministic pipelines + LLM Agent, precise line-level comments, built-in...

    Get the code, then follow its README to add the skill: git clone https://github.com/alibaba/open-code-review

  8. DesktopCommanderMCP

    MCP server · GitHub · 9,140 stars

    This is MCP server for Claude that gives it terminal control, file system search and diff file editing capabilities

    Add it to Claude Code (from the official MCP Registry): claude mcp add desktop-commander -- npx -y @wonderwhy-er/desktop-commander

  9. git-mcp

    MCP server · GitHub · 8,046 stars

    Put an end to code hallucinations! GitMCP is a free, open-source, remote MCP server for any GitHub project

    Get the code, then follow its README: git clone https://github.com/idosal/git-mcp

  10. anti-slop

    Agent skill · GitHub · 5,111 stars

    Filters out generic AI-generated UI, text, and code. https://t.co/qXBnOUmr32 https://t.co/YynKw1cGl4 https://github.com/miqdadbadjuber/anti-slop

    Get the code, then follow its README to add the skill: git clone https://github.com/miqdadbadjuber/anti-slop

  11. sandbox

    MCP server · GitHub · 4,662 stars

    All-in-One Sandbox for AI Agents that combines Browser, Shell, File, MCP and VSCode Server in a single Docker container.

    Get the code, then follow its README: git clone https://github.com/agent-infra/sandbox

  12. kubefwd

    Open-source tool · GitHub · 4,105 stars

    Bulk port forwarding Kubernetes services for local development.

    Get the code, then follow its README: git clone https://github.com/txn2/kubefwd

Questions

What is the difference between an MCP server and a skill for developers?

An MCP server gives the assistant a new ability, such as reading GitHub issues or driving a browser. A skill gives it a way of working, such as how your team writes commit messages or reviews code. Most setups use both.

Do these work outside Claude Code?

MCP is an open protocol, so most servers also work in Cursor, VS Code, Windsurf and other clients that support it. Skills written as SKILL.md folders are read by Claude Code and a growing number of other agents.

Related

Your own list

This page ranks by popularity. SkillScout's scan ranks by fit: the AI you already use writes a short profile of how you work, and SkillScout scores about 22,000 AI tools against your gaps and hands back a short list with install steps. It is free, and the server never saves your profile.

Run your scan