Which MCP Servers Connect to Productivity Platforms?

Productivity servers connect AI agents to the platforms where teams organize knowledge, publish content, track tasks, and schedule time. Independent evaluations of every significant server in this category, covering Notion, WordPress, Obsidian, Todoist, Google Calendar, and Linear, not directory listings or vendor documentation.

Model Context Protocol (MCP) lets AI clients create pages, update databases, manage tasks, and schedule events across productivity platforms through a standardized JSON-RPC 2.0 interface. Anthropic introduced MCP as an open standard in November 2024.

The Linux Foundation’s Agentic AI Foundation (AAIF) now governs the specification (Linux Foundation, “Agentic AI Foundation Launch,” December 9, 2025). 6 servers compete in this category as of mid-2026. 3 provide official vendor-maintained implementations.

The full MCP server directory covers all 14 server categories. This page focuses on productivity servers alone.

What Does a Productivity MCP Server Do?

A productivity MCP server exposes knowledge management, content publishing, task tracking, or calendar operations as callable tools. The AI client creates pages, updates tasks, schedules events, and searches workspaces through a standardized JSON-RPC interface. The server processes authentication and permission scoping.

Productivity servers bridge the model to the platforms teams already use. The agent reads a Notion database, creates a WordPress draft, checks a Todoist project, or schedules a Google Calendar event without the user switching applications.

4 platform types define this category:

• Knowledge management servers connect to Notion, Obsidian, and similar platforms. Tools expose page creation, database queries, content editing, and workspace search. Data models vary: Notion uses structured databases with properties. Obsidian uses local markdown files with wiki-style links.
• Content management servers connect to WordPress and similar publishing platforms. Tools expose post creation, media upload, page management, and plugin control.
• Task management servers connect to Todoist, Linear, and similar project tracking platforms. Tools expose task creation, status updates, priority assignment, and sprint planning.
• Calendar and scheduling servers connect to Google Calendar and similar platforms. Tools expose event creation, availability checking, and calendar listing.

Each platform type carries a different data architecture. Knowledge management platforms store structured and unstructured content. Task platforms store discrete items with statuses and priorities. Calendar platforms store time-bound events. The data architecture determines what operations the agent can perform.

How Do You Evaluate Productivity Servers?

Evaluate productivity servers on 4 criteria: platform type, data architecture (hosted cloud vs local-first), vendor maintenance status, and tool surface breadth. These dimensions determine whether a server fits the team’s existing workflow.

Platform Type

Knowledge management servers provide deep workspace integration. Content management servers provide publishing workflows. Task servers provide item-level CRUD operations. Calendar servers provide scheduling. Match the server to the platform the team already uses daily.

Data Architecture

Notion MCP connects to cloud-hosted databases through OAuth or API tokens. Obsidian MCP processes local markdown files on the user’s machine. WordPress.com provides native MCP without a plugin. The architecture determines where data lives and who controls access.

Vendor Maintenance Status

Notion, WordPress, and Google Calendar provide official vendor-maintained MCP servers. Obsidian, Todoist, and Linear rely on community implementations. Official servers receive API updates, security patches, and enterprise features. Community servers depend on volunteer maintenance.

Tool Surface Breadth

Notion MCP exposes 22+ tools. Obsidian community servers expose 25+ tools in advanced implementations. WordPress MCP covers posts, pages, media, users, plugins, and themes. Todoist MCP covers tasks, projects, labels, sections, and comments. Broader tool surfaces consume more context window tokens at session start.

The rubric above applies to every server profiled in the next section.

Which Servers Connect to Productivity Platforms Through MCP?

6 productivity servers expose MCP-compatible tools as of mid-2026. Notion, WordPress, and Google Calendar provide official vendor-maintained servers. Obsidian, Todoist, and Linear rely on community implementations. Notion leads the category with 4,200+ GitHub stars.

Notion MCP (Official)

Notion maintains the official Notion MCP server. The server has 4,200+ GitHub stars, 539 forks, and approximately 1.3 million all-time PulseMCP visitors (#46 globally). The current npm version is v2.2.1, targeting Notion API version 2026-03-11.

Notion provides 2 server implementations. The remote server at mcp.notion.com uses OAuth authentication and requires zero local setup. The local server (@notionhq/notion-mcp-server on npm) uses integration tokens and runs on the user’s machine. Notion is sunsetting the local server. The remote server is where active development and support are concentrated.

The server exposes 22+ tools organized across 4 workflow areas. Core operations include workspace search, page reading with stable block IDs, page creation, and database querying. Editing operations include markdown content editing, content appending, comment management, and user access. Optional MCP Apps provide interactive workflows (Data Source Explorer, Page Workbench).

Security-conscious users configure read-only integration tokens. Write access is optional and configurable per integration. The server intentionally omits destructive operations (database deletion is not exposed through MCP).

WordPress MCP (Official)

WordPress provides official MCP support through 2 paths. WordPress.com embedded native MCP in early 2026. No plugin installation required. Self-hosted WordPress installations use the WordPress/mcp-adapter package.

The original Automattic/wordpress-mcp plugin was archived on GitHub on January 19, 2026. The mcp-adapter replacement provides tools for managing posts, pages, media, users, plugins, and themes. WordPress.com’s native integration is the simplest path: MCP support activates without any configuration step.

Obsidian MCP (Community)

Community-maintained Obsidian MCP servers connect to local vaults through the Obsidian Local REST API plugin. The architecture is local-first: every note, image, and link graph processes on the user’s machine. No data leaves the vault.

Advanced implementations expose 25+ tools with capabilities no cloud-based server matches. Image vision reads visual content inside vaults. Link graph traversal identifies orphaned notes and broken connections. Section-level editing modifies content under specific headings without overwriting the entire page.

Auth: Obsidian API key configured locally. Transport: stdio via desktop proxy tunnel. No npm installation required in some implementations.

Todoist MCP (Community)

Community-maintained Todoist MCP servers connect to the Todoist API for task management. Doist (the company behind Todoist) provides an official package (@Doist/todoist-mcp). Tools expose task creation, retrieval, updating, moving, closing, reopening, and deletion. Project management, labels, sections, comments, and priority assignment are supported.

Auth: Todoist API key. Natural language input is supported for task creation with deadlines, priorities, and labels.

Google Calendar MCP (Official)

Google maintains the official Google Calendar MCP server as part of the Google Workspace suite. The server is remote hosted at calendarmcp.googleapis.com. Tools expose calendar listing, event creation, event updating, event deletion, and availability checking.

Auth: OAuth 2.0 through a Google Cloud project. The Google Workspace suite separates Calendar, Gmail, Drive, People, and Chat into 5 independent servers. Each authenticates through the same Google Cloud project. Agents install only the productivity tools they need.

Linear MCP (Community)

Community-maintained Linear MCP servers connect to the Linear API for developer-focused project management. Tools expose issue creation, status updates, sprint planning, priority assignment, and workspace summaries. Linear MCP targets engineering teams that use Linear for issue tracking alongside GitHub or GitLab for code. The server pairs well with GitHub MCP for workflows that span both issue tracking and pull request management.

Server selection depends on the platform. Notion MCP covers knowledge management. WordPress MCP covers content publishing. Todoist and Linear cover task tracking. Google Calendar covers scheduling.

The sections above define productivity MCP servers, provide an evaluation rubric, and profile every significant server in the category. The sections below compare hosted and local deployment models and answer the questions that surface during platform selection.

When Does a Hosted Server Outperform a Local Server?

Hosted servers require zero local setup and receive automatic API updates. Local servers process data on the user’s machine with full privacy. Notion is sunsetting its local server in favor of the hosted remote version. The trend favors hosted deployment for productivity platforms.

Hosted servers (Notion remote, Google Calendar, WordPress.com native) connect through a remote URL. OAuth handles authentication. API updates deploy server-side without user action. The tradeoff: data transmits to the vendor’s infrastructure. The user trusts the platform’s existing data handling policies.

Local servers (Obsidian MCP, Notion local) run on the user’s hardware. Data stays on the machine. No network dependency for core operations. The tradeoff: the user manages installation, updates, and API compatibility.

Notion’s decision to sunset the local server signals the ecosystem’s direction. Productivity platforms are consolidating around hosted MCP endpoints with OAuth authentication. Obsidian remains the exception: its local-first architecture is a design principle, not a temporary state.

Frequently Asked Questions

Which Productivity Server Is Most Installed?

Notion MCP leads with 4,200+ GitHub stars and approximately 1.3 million all-time PulseMCP visitors. Google Calendar MCP benefits from the broader Google Workspace suite adoption. Todoist MCP has multiple implementations across npm.

Can One Server Cover Notes, Tasks, and Calendar?

No single productivity MCP server covers all 3 functions. Notion comes closest by combining pages, databases, and task-tracking capabilities in one platform. Calendar functionality requires a separate server (Google Calendar MCP). Most productivity workflows install 2 to 3 servers: one for knowledge management, one for task tracking, and one for scheduling.

What Happened to the Original WordPress MCP Plugin?

Automattic archived the original wordpress-mcp plugin on GitHub on January 19, 2026. The replacement is WordPress/mcp-adapter for self-hosted installations. WordPress.com sites received native MCP support in early 2026 without requiring any plugin. Outdated guides referencing the archived plugin still appear in search results. Use the mcp-adapter package or WordPress.com native support.

Does Obsidian MCP Send Data to the Cloud?

No. Obsidian MCP processes every operation locally through the Obsidian Local REST API plugin. Notes, images, and the link graph remain on the user’s machine. No cloud service is involved. The local-first architecture is a deliberate design choice that differentiates Obsidian from cloud-based alternatives.

How Many Productivity Servers Can You Run Simultaneously?

Most productivity workflows run 2 to 3 servers. A common stack combines Notion MCP for knowledge management, Todoist MCP for task tracking, and Google Calendar MCP for scheduling. The combined tool definitions consume approximately 3,000 to 5,000 tokens at session start. Adding a 4th server (Linear or WordPress) is practical but increases context overhead. Each additional server’s tool definitions compete for space in the model’s context window.