MCP servers are quickly becoming the way AI assistants connect to the tools you actually use. Instead of copying data back and forth between your CRM and your AI chat, an MCP server gives the AI direct access - it can read your data, answer questions, and in some cases take action.

I have been testing every MCP server I can find that is relevant to RevOps, and I started maintaining a public list on GitHub. It just hit its first few stars, which is exciting for what started as a personal reference.

Here is a breakdown of what is out there and what is actually useful.

CRM & Sales

This is where the most impactful servers live, because CRM data is where RevOps teams spend most of their time.

HubSpot Remote MCP Server - HubSpot’s own first-party server, generally available since 2026-04-13 at https://mcp.hubspot.com. Free across all hubs and tiers. OAuth 2.1 + PKCE. Supports read and write on 12 standard CRM object types (contacts, companies, deals, tickets, carts, products, orders, line items, invoices, quotes, subscriptions, segments) and all 5 engagement types (calls, emails, meetings, notes, tasks), plus read-only access to organizational context and marketing content. Custom object schemas are not exposed today - that’s the clearest capability gap, and one of the main reasons RevOps teams still reach for third-party MCPs.

daeda-mcp - an open-source npm package (free) that takes a different architectural approach: it syncs contacts, companies, and deals to an encrypted SQLite database on your own machine via a HubSpot private-app token, so your AI assistant can run read-only SQL against a local mirror without network round-trips. Strictly scoped to those three objects. Good for portfolio-wide analysis from Claude Desktop or Cursor without chewing through API rate limits. For writes, custom objects, multi-portal, or Human-Approved Write Plans, Daeda AI’s Automation Suite adds those capabilities on top.

Automation & Analytics

MCP servers for tools like PostHog, Zapier, and workflow platforms let your AI assistant understand what is happening across your automation stack. Instead of switching between dashboards, you can ask questions like “which workflow had the most failures this week?” or “what is our funnel conversion rate for Q1?”

Project Management & Documentation

Servers for tools like ClickUp, Linear, and Notion let your AI assistant pull context from where your team actually works. When you are planning a HubSpot implementation, having your AI read the project brief from your PM tool and reference it while building the system is a genuine time saver.

What Makes a Good MCP Server

Not all MCP servers are equally useful. The ones that work best for RevOps share a few traits:

• Local data access - servers that sync data locally so queries are fast and do not burn API rate limits
• Write capabilities - read-only is useful for analysis, but the real power comes when the AI can also take action
• Granular permissions - you want control over what the AI can and cannot do, especially when write access is involved
• Real tool depth - a server that exposes two endpoints is less useful than one that gives comprehensive access to the platform

The List

The full list is public and open for contributions on GitHub. It covers CRM, automation, analytics, PM, documentation, and more - organised by category with notes on what each server actually does well.

If there is a server you are using that I have not included, I would love for you to add it. The list gets better when it reflects what real teams are actually finding useful.

Browse the Awesome MCPs list: The List

Try Daeda Portal Insights: NPM js Package