Skip to main content
Resources · MCP Integration

Set up bestdesq as an MCP server

The Model Context Protocol (MCP) makes bestdesq machine-readable and actionable for AI clients. This page shows how to connect your AI agent to your bestdesq account – and what it can do once connected.

Setup

1. Create a personal API token

In bestdesq, open Profile → Settings → API Tokens and click + Token.

  • Name: a descriptive label, e.g. Claude Code MacBook
  • Permissions: full access (recommended for your own trusted clients) or select individual scopes
  • Expiry: 30 days / 90 days / 1 year / No expiry

The token is shown once in plain text. Directly below the token a ready-to-use JSON snippet appears for your MCP client configuration – copy it with one click.

2. Configure your MCP client

The MCP endpoint for your bestdesq account is:

https://[your-account].bestdesq.app/mcp

Claude Code~/.config/claude-code/.mcp.json or project-local .mcp.json:

json
{
  "mcpServers": {
    "bestdesq": {
      "type": "http",
      "url": "https://[your-account].bestdesq.app/mcp",
      "headers": {
        "Authorization": "Bearer bdq_pat_YOUR_TOKEN_HERE"
      }
    }
  }
}

Cursor – enter the same structure under Settings → MCP. Cursor detects the server automatically on the next chat start.

Restart your client after saving. Once the connection is established, the agent will confirm it – or just ask: "Connect to bestdesq."

The bestdesq skill is a text file that explains to your AI agent how bestdesq works – workflows, conventions, when to act and when to ask. Without the skill, agents can use the tools but have to improvise more.

You set up the skill once in your AI environment. It is a plain text file and can be adapted to your own workflows.


Available tools

bestdesq provides 32 MCP tools, grouped by area. Which tools a token can see depends on the configured permissions – a read-only token won't even see bestdesq_create_task.

Projects & overview

ToolWhat it does
bestdesq_list_projectsList all projects, filterable by status
bestdesq_get_project_summaryProject details with task statistics, budget and time utilisation
bestdesq_get_dashboardPersonal overview: open tasks, deadlines, project health, due invoices
bestdesq_get_status_keysQuery valid status values for a project workflow

Typical requests:

"Which projects are active and where are there budget issues?"
"Give me my daily overview with all open tasks and due invoices."

Tasks

ToolWhat it does
bestdesq_list_tasksList tasks in a project, filterable by status and priority
bestdesq_get_taskTask details including comments and checklists
bestdesq_search_tasksFull-text search across task titles
bestdesq_find_overdue_tasksFind overdue tasks
bestdesq_create_taskCreate a new task
bestdesq_update_taskUpdate a task – status, title, priority, assignment
bestdesq_create_task_headingCreate a heading / group in the project
bestdesq_add_commentAdd a comment to a task
bestdesq_add_checklistCreate a checklist group on a task
bestdesq_add_checklist_itemsAdd items to an existing checklist
bestdesq_toggle_checklist_itemCheck or uncheck an individual item

Typical requests:

"Create a task 'Revise hero section' with high priority in the Website Relaunch project."
"Which tasks in project 26-004 are overdue and who are they assigned to?"
"Mark task 1234 as done."

Time tracking

ToolWhat it does
bestdesq_log_timeBook time on a task
bestdesq_get_time_summaryTime report by project, user or period

Typical requests:

"Book 90 minutes on task 'Wireframes' in project 26-002, today, description: concept revised."
"How much time did the team book on the TMV portal in April?"

Outgoing invoices

ToolWhat it does
bestdesq_list_planned_invoicesList planned invoices
bestdesq_get_invoicing_statusOverview: due, overdue, recently created
bestdesq_get_invoiceRead a single invoice
bestdesq_generate_invoiceConvert a planned invoice into a real invoice
bestdesq_update_planned_invoiceEdit a planned invoice

Typical requests:

"Which invoices are due this month?"
"Create the advance invoice for project Müller GmbH, service period May 2026."

Incoming invoices

ToolWhat it does
bestdesq_fetch_incoming_invoicesFetch new invoices from the email inbox; e-invoices are parsed automatically
bestdesq_parse_incoming_invoiceParse a PDF or XML file and extract document data
bestdesq_list_incoming_invoicesList saved incoming invoices
bestdesq_get_incoming_invoiceRead a single incoming invoice
bestdesq_import_incoming_invoicesSave parsed documents as incoming invoices
bestdesq_update_incoming_invoiceEdit a saved incoming invoice
bestdesq_upload_incoming_invoice_attachmentLink an attachment to an incoming invoice

Typical requests:

"Fetch new incoming invoices from the inbox and show me the extracted amounts."
"Import the parsed invoices – the supplier is already assigned."

Contacts

ToolWhat it does
bestdesq_list_organisationsList customers and suppliers
bestdesq_search_organisationsSearch by name, abbreviation or VAT ID
bestdesq_create_organisationCreate a new organisation

Typical requests:

"Is there already a supplier entry for Muster AG?"
"Create a new customer entry for Digital Studios GmbH."


Permissions

Every API token has scopes that determine which tools it may use. Scopes follow the pattern entity:action – a token with task:write implicitly also has read access.

The effective access is always the intersection of the token scopes and the user's role. A token can never do more than the user would be able to see in the bestdesq UI.

Recommended scope profiles:

Use caseScopes
Read and research onlyproject:read, task:read, time-record:read
Task work and time trackingproject:read, task:write, time-record:write
Outgoing invoicesadditionally invoice:read, invoice:write
Incoming invoicesincoming-invoice:read, incoming-invoice:write, organisation:read
Full access (own trusted clients)no scopes selected

What the agent does not do

  • Invoices are not sent. The agent can create invoices – sending them is up to you.
  • The agent acts within your token's scope. What your token is not allowed to do, the agent won't even see as an available tool.
  • No automatic actions. The agent only executes tool calls on explicit request – it does not monitor in the background.

Further reading