Connect Proyex to the AI Assistant (MCP)

Connect Proyex to AI assistants like Claude, ChatGPT, or Gemini using the Model Context Protocol (MCP). Once set up, your AI assistant can directly access your projects, tasks, milestones, files, comments, and CRM data—reading, creating, editing, and deleting them.

Written By Miro Flemke

Last updated About 1 month ago

You retain full control: Each connection is individually authorized via OAuth and can be revoked at any time in your profile settings.

What is MCP?

MCP (Model Context Protocol) is an open standard that allows AI assistants to access external tools and data directly. When you connect Proyex via MCP, your AI assistant can view and edit your projects, tasks, CRM data, and more—right from within the chat.

Supported AI Assistants

  • Claude (claude.ai and desktop app)

  • ChatGPT

  • Google Gemini

  • Cursor

  • Windsurf

  • Other MCP-compatible clients

Setup

Select Team

All actions in Proyex are team-based. Your AI assistant can use the "Show my teams" command to find out which teams you belong to and then work specifically within the correct team.

Claude

  1. Open claude.ai and go to Settings → Connectors

  2. Click “Add Custom Connector”

  3. Enter the following:

    • Name: Proyex

    • Remote MCP Server URL: https://app.proyex.io/mcp

    • Leave the OAuth Client ID and Secret fields blank

  4. Click Save

  5. The first time you make a call, you’ll be automatically redirected to the Proyex login page

  6. Log in and confirm the permissions—done!

ChatGPT / Gemini

  1. Open your AI assistant’s settings

  2. Look for the section for MCP servers or connectors

  3. Add the server URL: https://app.proyex.io/mcp

  4. Confirm the OAuth login when prompted

Cursor / Windsurf (IDE)

  1. Open the MCP settings in your IDE

  2. Add a new MCP server with the URL: https://app.proyex.io/mcp

  3. Complete the OAuth login in your browser

  4. Proyex tools are now available to you in the IDE’s AI Assistant

What can you do with MCP?

In total, your AI Assistant has 44 tools across 8 categories:

Projects

List projects, view details, create, edit, and delete projects, and view statistics.

Tasks & Subtasks

List, create, edit, and delete tasks and subtasks. View task details, including status, assignment, and due date.

Milestones

List, create, edit, mark as completed, and delete milestones.

Comments

View, create, edit, and delete comments on tasks.

Files

List and delete project files.

Time Tracking

View time entries and create new ones.

CRM (Deals, Companies, Contacts, Activities)

Manage the entire sales process: list, create, edit, and delete deals. Maintain companies and contacts. Log activities such as calls, meetings, and follow-ups.

Team

List team members and view team activities.

Resources — Data Overviews for the AI Assistant

In addition to the tools, your AI assistant can access 4 resources that provide structured overviews:

  • Team Overview — Number of members, number of projects, active projects

  • Project Summary — Progress, task status, milestones, overdue tasks

  • Task Board — Overview of a project’s tasks

  • CRM Pipeline — Deal status and sales metrics

AI Prompts — ready-made analysis templates

For complex analyses, there are 3 pre-built prompts available for the AI assistant to use:

  • Project Status Report — Generates a comprehensive report covering progress, milestones, roadblocks, and recommendations

  • Task Planning — Breaks down goals into specific tasks with assigned milestones, responsible parties, and deadlines

  • CRM Deal Analysis — Analyzes pipeline health, identifies at-risk deals, and calculates weighted revenue forecasts

Sample Questions for the AI Assistant

  • “Show me all open tasks in the Website Relaunch project”

  • “Create a new task: Revise header design, due Friday”

  • “What’s the current status of the App Development project?”

  • “Which deals are in the pipeline, and what is their total value?”

  • “Log 2 hours for the ‘Marketing Campaign’ project”

  • “Create a status report for the Onboarding project”

  • “Which deals are at risk? Analyze my pipeline”

  • “Plan the next steps for the goal: Launch the landing page”

Frequently Asked Questions

Do I need a specific plan for MCP?

Yes, MCP access is available starting with the Startup Plan.

Can I restrict access?

Yes, during OAuth authorization, you select the permissions (scopes). For example, you can allow read-only access.

Does MCP also work in Claude’s desktop app?

Yes, setup works in both the web version and the Claude desktop app.

What happens if I change my Proyex password?

The MCP connection remains active because it runs via OAuth tokens and isn’t tied to your password.