Skip to content
Documentation

MCP connector

Updated 28 October 2026

/agent-integrations/mcp-connector

MCP connector

The MCP connector brings an organisation's TechWolf data into the AI tool it already uses. People ask Claude, ChatGPT or an agent built in Microsoft Copilot Studio about the organisation's skills, jobs and tasks, and the AI tool answers from that data. The connector is TechWolf's server for the Model Context Protocol (MCP), the open standard that lets AI tools read data from other systems.

Key attributes

Read-only
Nothing asked through the connector changes the organisation's data in TechWolf.
Answers follow each person's Console access
Each person connects their own TechWolf account, and the AI tool sees only what that person can see in the TechWolf Console.
Works alongside other connectors
With the MCP connectors of HR, finance or planning systems added too, the AI tool can combine their data with TechWolf data in one answer.
No TechWolf credits
Questions use no TechWolf credits. Market and export queries have fair-use limits per minute, per person and per organisation, and the connector tells the AI tool when to try again. The AI tool's usage counts towards the organisation's plan with its provider.

How it works

The AI tool does the reasoning. The connector serves TechWolf data and runs no AI model; the AI tool decides what to read and writes the answer, so wording and figures can vary between tools and attempts.

The connector reads the data below. In the example questions, the part in brackets stands for something from the organisation.

TopicWhat the connector readsExample question
At a glanceHow many employees, jobs and job families TechWolf holds"Give me an overview of what TechWolf holds for our organisation."
EmployeesSkill profiles inferred from each person's work"How many people in [organisational unit] hold [skill]?"
JobsJob families, jobs and the skills each job needs"Which skills does the [job title] job need?"
MobilityHow people's skills match a job or an open vacancy"Who could move into the [job title] job, and what would they need to learn? Counts only, no names."
TasksEach job's tasks, how AI changes them, and jobs with similar work"List every task of the [job title] job with its AI impact level and the reason behind it."
TeamsOrganisational units and how they nest"Which teams sit under [organisational unit]?"
LearningCourses and the skills they teach"Which courses in our catalogue teach [skill]?"
MarketPublic job postings"Which skills do job postings for [job title] in [country] ask for most?"
  • Employee names or IDs. Depending on the data the organisation shares with TechWolf, answers show employees by name or by employee ID.
  • Data handling. The connector runs in the TechWolf region that holds the organisation's data and reaches only that TechWolf data, not what else the AI tool can see, such as Microsoft 365 files or email. Security and data handling covers the rest of an IT and security review.

Prerequisites

  • TechWolf data. The connector answers from whichever of these the organisation has:
    • Skills Intelligence for skills questions, such as a skills inventory or internal mobility.
    • Work Intelligence, on top of Skills Intelligence, for tasks, work redesign and AI impact.
    • Market Intelligence for labour-market questions. Every organisation has it.
  • A TechWolf Console account for each person, with the TechWolf Agent role and access to the data their questions need. People sign in as they do to the Console.
  • A supported AI tool that the organisation manages, where an admin can add a custom connector.

Availability

Every TechWolf organisation includes one MCP connector, which TechWolf turns on for each environment the organisation uses.

Setting up takes four parties, each with a guide people can open after they sign in:

  1. TechWolf turns on the connector and sends the TechWolf Console admin its address.
  2. The TechWolf Console admin gives people access, with Give people access to the Analyst Agent and the MCP connector.
  3. The AI tool admin adds TechWolf once for everyone, with the Admin setup guide for the tool. In Copilot Studio, a Microsoft 365 admin also approves an agent offered to everyone. Adding it takes minutes, but the security review usually takes longer, so it starts first.
  4. Each person connects their own TechWolf account, with the User setup guide for their tool.