DEV Community: curatedmcp

MCP Ecosystem Week 17: Official Integrations Dominate as Ecosystem Reaches 64 Verified Servers

curatedmcp — Mon, 20 Apr 2026 10:52:07 +0000

Originally published at curatedmcp.com/blog/week-2026-17

MCP Ecosystem Week 17: Official Integrations Dominate as Ecosystem Reaches 64 Verified Servers

The MCP marketplace is hitting a natural inflection point. With zero new servers added this week, the focus has shifted from rapid catalog expansion to deepening integrations with the tools developers already use daily. The data tells a clear story: official integrations from major platforms are capturing mindshare and usage, while the broader ecosystem consolidates around proven patterns.

This Week in MCP

No new servers were added to the marketplace this week, but that's not a sign of stagnation—it's a sign of maturation. The catalog stands at 64 verified servers, all free to use, and the community is channeling energy into building on top of existing infrastructure rather than fragmenting it further.

What is new this week: 8 fresh playbooks hit the platform. These curated workflows bridge the gap between raw MCP capabilities and real-world developer tasks, showing how to compose multiple servers into coherent agent behaviors. Playbooks are becoming the connective tissue of the ecosystem—a way to demonstrate value without requiring new server implementations.

Trending Servers

1. GitHub Copilot MCP (98,000 views) leads by a wide margin, reflecting the reality that code intelligence is table stakes for AI-assisted development. This one integrates Copilot's completions and code review capabilities into any MCP client, letting developers lean on GitHub's models and reasoning outside the VS Code sandbox.

2. OpenAI MCP (87,000 views) sits right behind, unsurprising given OpenAI's reach. The ability to orchestrate GPT-4o, DALL-E, Whisper, and embeddings through a single protocol makes this essential infrastructure for teams already deep in the OpenAI ecosystem.

3. Figma MCP (82,000 views) rounds out the top three and signals something important: it's not just about code anymore. Design-to-development workflows are a legitimate use case for AI agents, and Figma's official server taps into a large developer audience that treats design files as first-class development assets.

Also worth noting: GitHub MCP and Anthropic Claude MCP both pulled strong numbers (76K views each). The former is obvious (repo management from an agent context is powerful), but the latter deserves attention—nesting Claude within Claude for specialized reasoning tasks suggests developers are starting to think about agentic decomposition and task-specific sub-agents as a legitimate architecture pattern.

Quick Take

The top 5 trending servers are all either officially maintained by major platforms (GitHub, OpenAI, Figma, Anthropic) or deeply integrated with platform ecosystems (Copilot). This concentration around official integrations isn't surprising, but it does highlight a constraint: developer trust and adoption clustering around names they already recognize.

The real opportunity for community-built MCP servers likely lies in specialized domains—internal tools, niche SaaS integrations, domain-specific APIs—where official support doesn't exist. The 100% free pricing model helps here: there's no friction to adoption once a server proves useful, but discoverability and trust remain the gatekeepers. The 8 new playbooks this week suggest the community understands this, focusing on composition and narrative rather than net-new tools.

Browse all 64 verified servers at CuratedMCP.

GitHub MCP: Manage Your Entire Repo Workflow from Your AI Agent

curatedmcp — Mon, 20 Apr 2026 10:51:59 +0000

Install guide and config at curatedmcp.com

GitHub MCP: Manage Your Entire Repo Workflow from Your AI Agent

GitHub MCP is the official Model Context Protocol server that connects your AI agent directly to GitHub. Instead of switching between your editor, GitHub's web UI, and CI/CD dashboards, you can now handle pull requests, issues, workflows, and repository management through natural language commands in Claude, Cursor, or Windsurf.

What It Does

This server gives AI agents full visibility into your repositories and the ability to take action on them. Your agent can review pull requests with full code context, search and triage issues across multiple repos, query commit history and file changes, manage branch protections and collaborators, and trigger—or inspect—GitHub Actions workflows without leaving your development environment.

The practical upside: when you ask Claude to "review that PR and check if the tests passed," it actually opens the PR, reads the code, checks the CI logs, and gives you a detailed breakdown. You're not describing problems to your agent; your agent is seeing them directly.

Features include creating and merging PRs, searching issues and comments, managing collaborators, accessing security alerts and dependency insights, and fetching repository metadata for code browsing. It's the kind of integration that turns your AI agent from a chat interface into an actual development teammate.

How to Install

Install the MCP server via npm:

npx -y @github/mcp

Then add it to your Claude Desktop configuration:

{
  "mcpServers": {
    "github-mcp": {
      "command": "npx -y @github/mcp"
    }
  }
}

Authenticate with your GitHub token when prompted, and you're ready to go.

Real-World Use Cases

Automated PR Triage: Ask Claude to "find all open PRs with failing tests, summarize the failures, and suggest fixes." Your agent reads each PR, checks the workflow logs, and delivers actionable feedback in seconds.
Issue Investigation: "Search for all open bugs labeled 'critical' and create a summary of what needs fixing." The agent finds issues across repos, reads comments and linked PRs, and synthesizes context.
Workflow Debugging: When a GitHub Actions pipeline fails, ask your agent to "check the latest workflow run for my deploy job and show me what went wrong." It fetches the logs directly and explains the error.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.

Salesforce MCP: Give Your AI Agent Full CRM Access via Natural Language

curatedmcp — Sun, 19 Apr 2026 10:26:45 +0000

Install guide and config at curatedmcp.com

Salesforce MCP: Give Your AI Agent Full CRM Access via Natural Language

Salesforce just launched its official MCP (Model Context Protocol) server, and it's a game-changer if you're building AI agents that need to touch your CRM. Instead of building custom APIs or manual integrations, you can now drop this server into Claude, Cursor, or Windsurf and let your agent search records, create leads, run SOQL queries, and pull Einstein AI insights—all through conversation.

This is one of Anthropic's launch partners for MCP Apps. It renders interactive dashboards directly in Claude's chat window and handles OAuth authentication with granular Salesforce permission sets. Works with Enterprise, Unlimited, and Developer editions.

What It Does

The Salesforce MCP server gives your AI agent read and write access to your entire Salesforce org. Search and retrieve any object (Accounts, Contacts, Opportunities, Cases). Create or update records across standard and custom objects. Run complex SOQL queries without leaving your chat. Access Einstein AI predictions and insights. Manage your sales pipeline, handle approvals, check activity history, and pull reports and dashboards.

The real value: your agent can now understand your CRM's current state and act on it in real time. No more "let me check Salesforce and come back to you." It's all in one conversation window.

How to Install

npx -y @salesforce/mcp

Add it to your Claude Desktop config:

{
  "mcpServers": {
    "salesforce-mcp": {
      "command": "npx -y @salesforce/mcp"
    }
  }
}

You'll authenticate via OAuth—Salesforce handles permissions based on your user role.

Real-World Use Cases

Instant lead qualification: Ask Claude to search for high-value open opportunities, pull their account history, and suggest next steps based on activity patterns—without manually switching tabs.
Bulk deal updates: "Update all Q1 opportunities in the Healthcare vertical to the new forecast category and add a note." Your agent handles the SOQL query and batch updates.
Einstein-powered insights: Request competitor analysis or win/loss predictions tied to specific deals, then have your agent summarize findings and surface them in a dashboard.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.

Hugging Face MCP: Search 500k+ Models and Run Inference Directly From Claude

curatedmcp — Sat, 18 Apr 2026 10:26:45 +0000

Install guide and config at curatedmcp.com

Hugging Face MCP: Search 500k+ Models and Run Inference Directly From Claude

The Hugging Face MCP server connects your AI agent directly to the world's largest open-source model repository. Instead of leaving your Claude or Cursor session to browse Hugging Face, search models, run inference, explore datasets, and deploy Spaces without context-switching.

What It Does

This server unlocks five core capabilities:

Model discovery across 500,000+ models filterable by task, framework, and performance metrics. Find the right model for your use case without manual browsing.

Inference execution lets you run predictions through any model via the Hugging Face Inference API—pass text, images, or audio and get results back in your agent's context.

Dataset access to 150,000+ community datasets with schema inspection and preview data, useful for training, fine-tuning, or research workflows.

Space management to deploy and monitor Gradio/Streamlit applications directly from your agent.

Leaderboard comparisons so you can check Open LLM Leaderboard rankings and model benchmarks programmatically.

You get model cards, tokenizer configs, community discussions, and weight downloads all accessible without switching tabs.

How to Install

Install via npm:

npx -y @huggingface/mcp

Add to your Claude Desktop config:

{
  "mcpServers": {
    "hugging-face-mcp": {
      "command": "npx -y @huggingface/mcp"
    }
  }
}

After installation, restart Claude Desktop and the server will appear in your MCP options.

Real-World Use Cases

Model selection: Ask Claude, "Find me the best open-source sentiment analysis model under 1B parameters with the highest F1 score on the leaderboard." Get ranked results with links and stats instantly.
Inference pipelines: "Run this customer review through the Mistral-7B model and give me the sentiment score." Execute inference without writing API calls.
Data discovery: "What datasets are available for training a code completion model? Show me their sizes and schemas." Explore training data options while brainstorming architecture.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.

Exa MCP: Semantic Search for AI Agents That Actually Understands What You Mean

curatedmcp — Fri, 17 Apr 2026 14:14:12 +0000

Install guide and config at curatedmcp.com

Exa MCP: Semantic Search for AI Agents That Actually Understands What You Mean

Traditional search APIs return keyword matches. Exa MCP returns meaning — it's a neural search engine purpose-built for AI workflows. Instead of feeding Claude or Cursor fuzzy keyword queries, you get conceptually related results, even when exact terms don't overlap.

Think of it as the difference between searching for "machine learning classification" and getting papers about decision trees versus searching by idea and finding everything semantically similar, regardless of terminology. That's Exa.

What It Does

Exa MCP connects your AI agent to Exa's search API, unlocking capabilities that traditional search tools (like Brave or Tavily) can't match:

Semantic search — find content by meaning, not keywords
Find similar content — point to any URL and discover conceptually related pages
Research mode — synthesis across multiple sources with structured extraction
Temporal filtering — search before/after specific dates for trend analysis
Domain filtering — restrict to academic papers, news, GitHub repos, technical docs, etc.
Highlighted passages — extracts relevant snippets directly from results

This is particularly powerful for research-heavy workflows: finding academic papers similar to an abstract, discovering competitors doing X for a different vertical, or tracking what was written about a topic before a certain date.

How to Install

npx -y exa-mcp-server

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "exa-mcp": {
      "command": "npx -y exa-mcp-server"
    }
  }
}

Restart Claude Desktop, and Exa becomes available as a callable tool in your agent's context.

Real-World Use Cases

Competitive research — "Find companies building the Figma equivalent for 3D design" or "Show me startups using approach X differently." Exa finds conceptual matches, not just exact competitors.
Literature review — Hand Claude an arXiv paper abstract and ask it to "find similar research." Exa surfaces related work across the academic web without keyword matching.
Time-scoped discovery — "What were people writing about RAG before 2023?" Exa's date filtering lets you track how narratives evolved and find pre-hype historical context.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.

The Composable AI Agent

curatedmcp — Mon, 13 Apr 2026 15:19:38 +0000

Great article on the composable AI agent, how Model Context Protocol Is Redefining Software Infrastructure
https://www.curatedmcp.com/whitepaper#executive-summary

MCP Servers Just Got an App Store - And It's Already Changing How We Build AI Agents

curatedmcp — Sat, 11 Apr 2026 16:50:48 +0000

If you've spent any time building with Claude Desktop or Cursor lately, you've felt the pain of MCP server discovery.

You know the drill:

Step 1: Search GitHub for "mcp server stripe"

Step 2: Find 6 repos, 4 unmaintained

Step 3: Attempt claude_desktop_config.json wiring

Step 4: Debug for 3 hours

Step 5: Still unsure if this is safe to run

CuratedMCP just solved that.

It's a curated, human-reviewed catalogue of 34+ production-ready MCP servers — Stripe, Figma, AWS, Cloudflare, Vercel, Docker, and more — with one-click claude_desktop_config.json generation and an AI Stack Builder that takes a plain-English description of your goal and returns a complete server stack in under a minute.

For developers specifically: 80% revenue share. Five-minute publish flow. You own your category.

The MCP ecosystem is at the App Store moment of 2008. The developers who publish now will own the categories.

→ Publish your server | Browse the catalogue