MCP Integration — AI Assistant Setup

Connect Your AI Assistant
to Live Workforce Data

Connect Claude, ChatGPT, Gemini, Copilot, Cursor, or any MCP-compatible AI to TheDeskMonitor in minutes. Ask natural-language questions about your team — timesheets, clock-in status, workload alerts, productivity — and get live answers.

Get Started Free See 24 Live Tools
WORKS WITH: Claude Desktop Claude Code Cursor ChatGPT Gemini Copilot Perplexity 🌊 Windsurf
1
Get API key
Free demo or live key
2
Pick your AI client
Claude, ChatGPT, Cursor…
3
Paste config + restart
One JSON block, 2 min
4
Ask your first question
"Who is clocked in?"

Total setup time: under 5 minutes. Start below — no account required.

MCP STANDARD — LIVE IN PRODUCTION

What is MCP?

Model Context Protocol (MCP) is an open standard developed by Anthropic that lets AI assistants connect securely to external data sources and call real tools — without needing custom plugins built separately for each AI provider.

TheDeskMonitor ships a native MCP cloud server at api.thedeskmonitor.com/api/mcp/v1 with 24 live tools. Any MCP-compatible AI assistant can discover these tools automatically and answer workforce questions in natural language using your live data.

Setup takes under 5 minutes. No webhook URLs to manage. No third-party connector service. Your AI client talks directly to DeskMonitor's API over HTTPS.

How Authentication Works

Two modes — try it free before creating an account, or connect to your real tenant:

🧪
Demo Mode — No Account Needed
Use key dm_demo_public → explores a synthetic company (8 employees, 3 teams, 3 projects). Analytics + discovery tools active. No sign-up required.
🔑
Live Mode — Your Real Data
Sign in → Settings > API Keys → generate a dm_live_* key. Scoped to your tenant only. Supports clock-in/out, invite staff, and all 24 tools.

Data is tenant-scoped — your API key cannot access another organization's data. Employees see only their own records; managers see their team; admins see their tenant.

1

Get Your API Key

🧪
Demo Key — No Account
Free Instantly

Use this key in any MCP client config below to explore DeskMonitor with a synthetic 8-person company. No registration, no credit card.

dm_demo_public
🔑
Live Key — Your Real Data
Requires Account

Sign in to TheDeskMonitor, navigate to Settings > API Keys, and generate a key. Copy the dm_live_* key — you will use it in your AI client's config below.

Go to API Keys → Create Account
One-click installer — coming soon. The desktop connector package is not yet available for download. In the meantime, use the Manual Setup below — it takes under 5 minutes and works with Claude Desktop, Cursor, Windsurf, VS Code, and any MCP-compatible client today.
VISUAL WALKTHROUGH

What to Expect — Step by Step

Here is exactly what you will see at each step. Total time: under 5 minutes. No surprises — a consent screen will appear once when you first connect a live key.

Setup time: under 5 minutes from this page to asking your first workforce question.
Consent screen appears once: when you connect with a dm_live_* key, your AI client may trigger an authorisation screen — this is normal and expected. You review the permissions and click Allow. Demo mode (dm_demo_public) skips this step.
What success looks like: a tools icon (hammer 🔨) appears in your AI client's input bar. Type a question — the AI calls DeskMonitor's live tools and returns a real answer with names, hours, and status.
Write actions ask for confirmation: clock-in, clock-out, and invite-staff will always ask "Are you sure?" before doing anything — the AI will never act silently.
A
DeskMonitor Settings › API Keys page — showing the key list, a masked dm_live_ key, and the Generate New Key button
Get your API key
Sign in → Settings › API Keys → click Generate New Key → copy the dm_live_… value.
B
Claude logo claude_desktop_config.json
"mcpServers": {
  "deskmonitor": {
    "command": "npx",
    "args": [
      "mcp-remote@latest",
      "…/api/mcp/v1/",
      "Bearer dm_live_…"
    ]
  }
}
Paste key → save file → restart client
Paste the config into your AI client
Add the JSON block to your client's MCP config file (see the per-client tabs below).
C
DeskMonitor Authorise AI Access consent screen — showing Claude Desktop requesting read and write permissions, with Allow and Deny buttons
Review & Allow access live key only
A consent screen shows what the AI will be able to do. Review the permissions and click Allow. Only appears once per key.
D
Placeholder: screenshot of AI client showing the hammer / tools icon confirming DeskMonitor MCP tools are connected — to be captured from a running client
See the tools icon in your AI client
A 🔨 hammer icon (or "Tools" badge) appears in the chat input bar — this confirms DeskMonitor's 24 tools are active.
E
AI assistant answer showing live workforce data: 5 of 8 employees clocked in, names with clock-in times and hours worked, and a workload alert — returned via DeskMonitor MCP tools
Ask your first workforce question
Type "Who is clocked in?" — the AI calls DeskMonitor's live tools and returns real names, hours, and status in seconds.

Steps A–C use only our own pages (no client-side screenshots fabricated). Step D is a client-side capture — it will appear here once recorded from a running AI client. All data shown in Step E is from the free dm_demo_public key — illustrative only, not real employee data.

2

Configure Your AI Client

Select your AI assistant below and follow the numbered steps. Replace YOUR_KEY_HERE with your key from Step 1.

🤖

Claude Desktop

Live — Fully Supported Windows & macOS

Claude Desktop supports MCP natively via its claude_desktop_config.json file. It is the most seamless integration — Claude discovers all 24 DeskMonitor tools automatically and uses them when you ask workforce questions.

1 Open the Claude Desktop config file

The file location depends on your OS:

# Windows
%APPDATA%\Claude\claude_desktop_config.json

# macOS
~/Library/Application Support/Claude/claude_desktop_config.json

If the file does not exist yet, create it as an empty { } JSON file.

2 Add the DeskMonitor MCP server entry

Claude Desktop uses stdio transport and requires the mcp-remote bridge to reach a remote HTTP MCP server. The bridge is downloaded automatically by npx — Node.js v18+ must be installed.

{
  "mcpServers": {
    "deskmonitor": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote@latest",
        "https://api.thedeskmonitor.com/api/mcp/v1/",
        "--header",
        "Authorization: Bearer YOUR_KEY_HERE"
      ]
    }
  }
}

Replace YOUR_KEY_HERE with dm_demo_public for demo mode, or your dm_live_* key for live data. Why mcp-remote? Claude Desktop only supports stdio (local process) transport — it cannot connect directly to a remote HTTP MCP endpoint. mcp-remote acts as a local stdio-to-HTTPS bridge spawned as a child process; it handles all HTTP relaying and OAuth.

3 Restart Claude Desktop

Quit and reopen the Claude Desktop app. When the connection is active, you will see a tools icon (hammer) in the chat input bar, showing DeskMonitor's tools are available.

📸
Screenshot coming soon
Claude Desktop — tools icon (🔨) in chat input bar
Will show the hammer icon confirming DeskMonitor's 24 MCP tools are connected — captured from a running Claude Desktop instance.
4 Ask your first workforce question

Type any of these in the Claude Desktop chat to verify the connection is working:

"Who is clocked in right now?"
"Give me a morning briefing for my team."
"What's my DeskMonitor status?"
💻

Claude Code

Live — Fully Supported CLI · macOS, Windows, Linux

Claude Code is Anthropic's official CLI and supports MCP servers via the claude mcp add command or the shared config file. Once added, DeskMonitor tools are available in any Claude Code session — useful for developer workflows where you want workforce context alongside code assistance.

1 Add the MCP server via the CLI
# Add DeskMonitor MCP server (user-scoped, persists across sessions)
claude mcp add deskmonitor \
  --transport http \
  --url https://api.thedeskmonitor.com/api/mcp/v1/ \
  --header "Authorization: Bearer YOUR_KEY_HERE"
2 Or add manually to the MCP config file

Alternatively, edit ~/.claude/claude_desktop_config.json (same format as Claude Desktop above) to add the deskmonitor entry under mcpServers.

3 Verify the server is listed
claude mcp list
# Should show: deskmonitor http https://api.thedeskmonitor.com/api/mcp/v1/
4 Use DeskMonitor tools in any Claude Code session
# In Claude Code chat or REPL:
"Who is clocked in right now on my team?"
"Show me pending timesheet approvals."
"Clock me in." # Requires dm_live_* key
⚡

Cursor

Live — Fully Supported IDE · macOS, Windows, Linux

Cursor has built-in MCP support. Once DeskMonitor is added, you can query your team's status from within the IDE — useful for developers who want workforce context without switching apps.

1 One-click install — Add to Cursor

Click the button below to launch Cursor's built-in MCP installer. Cursor will open a dialog to confirm adding DeskMonitor — enter your API key when prompted.

⚡ Add to Cursor Opens Cursor directly — replace YOUR_KEY_HERE in the dialog

Prefer manual setup? Follow steps 2–3 below instead.

2 Or open Cursor Settings → MCP manually

In Cursor, press Ctrl+Shift+P (or Cmd+Shift+P on macOS) → type MCP → select Cursor Settings: Open MCP Config. This opens ~/.cursor/mcp.json.

3 Add the DeskMonitor server entry (manual config)
{
  "mcpServers": {
    "deskmonitor": {
      "url": "https://api.thedeskmonitor.com/api/mcp/v1/",
      "transport": "http",
      "headers": {
        "Authorization": "Bearer YOUR_KEY_HERE"
      }
    }
  }
}
4 Reload Cursor and verify the connection

Restart Cursor. In the Cursor Chat panel, you should see DeskMonitor listed under available tools. Ask a question like "Who is clocked in?" to verify.

📸
Screenshot coming soon
Cursor — MCP Settings panel listing DeskMonitor as a connected server
Will show the Cursor MCP panel with DeskMonitor listed and a green "connected" indicator — captured from a running Cursor IDE instance.
5 Ask workforce questions from within Cursor Chat
"Show me this week's team hours."
"Are there any workload alerts today?"
"What's my productivity score this week?"
💬

ChatGPT

Beta — Custom Connectors (MCP) ChatGPT Plus / Team / Enterprise

Beta status: OpenAI is rolling out MCP support via "Custom Connectors" (ChatGPT Plus/Team/Enterprise) as of mid-2025. Availability varies by account tier and region. The steps below describe the expected configuration once MCP Custom Connectors are available in your account. Check OpenAI's Help Center for current rollout status.

1 Open ChatGPT Settings → Connectors

In ChatGPT (web or desktop), go to Settings > Connectors (or the equivalent MCP section in your account). This option is available on ChatGPT Plus, Team, and Enterprise plans where the MCP beta has been enabled.

2 Add a new Custom Connector (MCP)

Select Add Connector (or Add MCP Server) and enter the following values:

Name: TheDeskMonitor
URL: https://api.thedeskmonitor.com/api/mcp/v1/
Auth: Bearer Token
Token: YOUR_KEY_HERE
3 Enable the connector in a Chat

Start a new chat and enable the DeskMonitor connector from the tools or connectors panel. ChatGPT will call tools/list to discover all 24 DeskMonitor tools automatically.

4 Ask your workforce questions
"Summarize my team's hours for this week using DeskMonitor."
"Who has pending timesheet approvals?"
📸
Screenshot coming soon
ChatGPT — Custom Connectors settings panel with TheDeskMonitor active
Will show the ChatGPT Connectors screen with TheDeskMonitor listed as an active MCP connector — available once ChatGPT's MCP Custom Connectors feature is generally available.
✨

Google Gemini

Live — MCP Supported Gemini CLI · Gemini Advanced

Status: Gemini CLI MCP is live and verified — 24 DeskMonitor tools are available via gemini mcp add. Gemini Advanced (web) MCP integration is on Google's roadmap but is not yet generally available. Follow the steps below to connect Gemini CLI now.

1 Gemini CLI — install and authenticate
npm install -g @google/gemini-cli
gemini auth login
2 Add DeskMonitor with one command

The Gemini CLI supports gemini mcp add to register an MCP server directly from the terminal — no config file editing needed:

gemini mcp add deskmonitor https://api.thedeskmonitor.com/api/mcp/v1/ \
  --header "Authorization: Bearer YOUR_KEY_HERE"

Or edit ~/.gemini/settings.json manually (see step 3 below). Refer to the Gemini CLI README if the mcp add subcommand is not yet in your installed version.

3 Or add manually to ~/.gemini/settings.json
{
  "mcpServers": {
    "deskmonitor": {
      "url": "https://api.thedeskmonitor.com/api/mcp/v1/",
      "transport": "http",
      "headers": {
        "Authorization": "Bearer YOUR_KEY_HERE"
      }
    }
  }
}
4 Use DeskMonitor tools in the Gemini CLI
gemini chat
> Who is clocked in on my team right now?

Gemini Advanced (web): When Google adds MCP connector support to Gemini Advanced, DeskMonitor will be compatible via the same HTTP MCP endpoint. We will publish a dedicated setup guide at that time. Contact us if you want to be notified when Gemini web support is ready.

🪟

Microsoft Copilot

Coming Soon — MCP Plugin Support Copilot for Microsoft 365 / Copilot Studio

Status: Microsoft has announced MCP support for Copilot Studio and GitHub Copilot (VS Code extension). Direct MCP connectivity for Copilot for Microsoft 365 is in development. The setup paths below describe the expected integration approach — exact steps will be confirmed when the feature reaches general availability.

1 GitHub Copilot (VS Code) — MCP Extension

GitHub Copilot in VS Code supports MCP tools in the Copilot Chat extension. Edit your VS Code settings.json or workspace .vscode/mcp.json:

// .vscode/mcp.json (workspace-scoped)
{
  "servers": {
    "deskmonitor": {
      "type": "http",
      "url": "https://api.thedeskmonitor.com/api/mcp/v1/",
      "headers": {
        "Authorization": "Bearer YOUR_KEY_HERE"
      }
    }
  }
}
2 Copilot Studio — Custom Connector

In Copilot Studio, create a Custom Connector using DeskMonitor's OpenAPI schema available at https://api.thedeskmonitor.com/api/mcp/v1/openapi.json. Set the connector authentication to API Key → Bearer Token with your DeskMonitor API key.

3 Ask workforce questions via Copilot Chat
"DeskMonitor: show me pending approvals."
"What is my team's productivity score this week?"
🔍

Perplexity

Coming Soon — MCP Actions Perplexity Pro

Status: Perplexity has announced MCP support for the Pro tier ("Perplexity Actions"). The feature is in phased rollout as of 2025. Once MCP Actions are available in your account, you will be able to connect DeskMonitor using the steps below.

1 Open Perplexity Settings → Actions (or MCP)

In your Perplexity Pro account, go to Settings > Actions (or the MCP section when available). Select Add Action or Add MCP Server.

2 Enter the DeskMonitor endpoint and API key
Endpoint: https://api.thedeskmonitor.com/api/mcp/v1/
Auth Type: Bearer Token
Token: YOUR_KEY_HERE
3 Enable DeskMonitor in a Perplexity search session

Enable the DeskMonitor action in your search session and ask a question. Perplexity will combine live DeskMonitor data with its web search results for comprehensive answers.

🌊

Windsurf

Live — Fully Supported IDE · macOS, Windows, Linux

Windsurf (by Codeium) has built-in MCP support. Once DeskMonitor is added, you can query live workforce data from within the Windsurf IDE — useful for agentic development workflows that need team context.

1 One-click install — Add to Windsurf

Click the button below to launch Windsurf's built-in MCP installer. Windsurf will open a dialog to confirm adding DeskMonitor.

🌊 Add to Windsurf Opens Windsurf directly — replace YOUR_KEY_HERE in the dialog

Prefer manual setup? Follow steps 2–3 below.

2 Or open Windsurf Settings → MCP manually

In Windsurf, press Ctrl+Shift+P (or Cmd+Shift+P) → type MCP → select Windsurf Settings: Open MCP Config. This opens ~/.codeium/windsurf/mcp_config.json.

3 Add the DeskMonitor server entry
{
  "mcpServers": {
    "deskmonitor": {
      "url": "https://api.thedeskmonitor.com/api/mcp/v1/",
      "transport": "http",
      "headers": {
        "Authorization": "Bearer YOUR_KEY_HERE"
      }
    }
  }
}

Save the file, then reload Windsurf. DeskMonitor tools will appear in the Windsurf Cascade MCP panel.

4 Ask workforce questions from Windsurf Cascade
"Who is clocked in right now?"
"Show me pending approvals for my team."
"What are the workload alerts today?"
🔌

Generic MCP Client

Live — Any MCP HTTP Client Continue.dev, Zed, custom integrations

Any MCP-compatible client that supports Streamable HTTP transport (MCP spec 2025-03-26) can connect to DeskMonitor. This includes Continue.dev, Zed editor, and any custom MCP integration you build.

1 Server details
MCP Endpoint: https://api.thedeskmonitor.com/api/mcp/v1/
Transport: Streamable HTTP (MCP spec 2025-03-26)
Auth: Authorization: Bearer YOUR_KEY_HERE
Tool discovery: GET https://api.thedeskmonitor.com/api/mcp/v1/tools
Health check: GET https://api.thedeskmonitor.com/api/mcp/v1/health
OpenAPI schema: GET https://api.thedeskmonitor.com/api/mcp/v1/openapi.json
2 JSON-RPC 2.0 example — call a tool directly
curl -X POST https://api.thedeskmonitor.com/api/mcp/v1/ \
  -H "Authorization: Bearer YOUR_KEY_HERE" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_dashboard_summary","arguments":{}}}'
3 Continue.dev setup

In ~/.continue/config.json, add DeskMonitor under the mcpServers array:

{
  "mcpServers": [
    {
      "name": "deskmonitor",
      "transport": { "type": "http", "url": "https://api.thedeskmonitor.com/api/mcp/v1/" },
      "requestOptions": { "headers": { "Authorization": "Bearer YOUR_KEY_HERE" } }
    }
  ]
}
3

Ask Your First Workforce Question

Once connected, try these questions in your AI client. They work with demo mode and live mode.

Attendance & Status
"Who is clocked in right now?"
"Am I clocked in?"
"Who hasn't clocked in yet today?"
Timesheets & Hours
"Show me my timesheet for this week."
"How many hours did my team work this month?"
"Who worked the most hours this week?"
Productivity
"What's my productivity score this week?"
"Are there any workload alerts today?"
"Give me a morning briefing for my team."
Approvals & Actions
"Show me timesheets waiting for approval."
"Invite james@acme.com to my workspace."
"Clock me in." (requires live key)

24 Live AI Workforce Tools

Every tool below is live and callable today from any connected AI assistant. All are available in demo mode. Write tools (clock in/out, invite staff) require a live API key.

get_dashboard_summary

Dashboard KPIs — who is in, hours today, productivity, pending items

Read
list_employees

Roster with clock-in status, department, and availability

Read
get_employee_details

Full profile — role, team, recent activity, hours

Read
list_my_timesheet

Your own time log entries for any date range

Read
summarize_team_hours

Aggregated team hours — per-person and totals

Read
list_pending_approvals

Timesheets awaiting manager sign-off

Read
list_recent_activity

Chronological feed — clock events, submissions, approvals

Read
get_workload_alerts

Anomaly flags — overtime risk, idle, late arrival, gaps

Read
get_productivity_overview

Active vs idle ratios and productivity scores

Read
clock_in

Start a work session for the authenticated user

Write
clock_out

End the current work session

Write
invite_staff

Send an email invitation to join your workspace

Write
get_capabilities

Role-scoped list of all available AI commands

Discovery
get_pricing_and_plans

Plans, features, and pricing details

Discovery
start_free_trial

Returns the sign-up URL and onboarding instructions

Discovery
get_started_guide

Onboarding guide and sample questions for first-time setup

Onboarding
get_data_dictionary

Schema reference — fields, metrics, and definitions

Onboarding
switch_to_demo

Switch the AI session to synthetic demo data

Demo
switch_to_production

Return from demo mode to your real tenant context

Demo
View Full Tool Reference →

Frequently Asked Questions

What is MCP and why does it matter for workforce management?
Can I try this without creating a DeskMonitor account?
What is the MCP endpoint URL?
Is my data safe? Who can see it?
Can my AI clock me in and out?
Does MCP cost extra?
Which AI clients fully support MCP today?
How do I revoke or rotate my API key?

All third-party product names and logos (Claude, ChatGPT, Cursor, Google Gemini, Microsoft Copilot, Perplexity) are trademarks of their respective owners. Use here indicates compatibility only and does not imply endorsement or affiliation.

Your AI assistant is ready for workforce intelligence.

Connect in under 5 minutes. Try demo mode free — no account required.

Get Started Free Create Account See all 24 tools