# MCP connector tools

The [MCP connector](/agent-integrations/mcp-connector) offers the AI tool a set of tools, each of which reads one kind of TechWolf data. People ask their question in their own words: the AI tool reads each tool's description, picks the tools the question needs and often calls several in turn. Admins see the tool names below when they add TechWolf to their AI tool.

## Key attributes

- **Read-only.** Every tool only reads data, and each is marked as read-only for the AI tool.
- **A reason with every call.** Every tool takes a `reason` input: one sentence the AI tool writes about the question it is answering. [MCP connector storage and logging](/agent-integrations/mcp-connector/security/storage-and-logging) describes the call record that keeps it.
- **The same tools for everyone.** Every person's AI tool lists the same tools, and each call checks that person's Console access. A call returns only the employees and jobs the person can see, and a tool that needs more access than the person has answers with the data access it needs.
- **Fair-use limits.** `query_exports` and `market_search` serve the export and market queries, which have fair-use limits per minute, per person and per organisation.

## How it works

The AI tool combines the tools. To answer "Which skills does the [job title] job need?", it looks the job up with `job_architecture`, which gives it the job's ID, then reads the job's skill profile with `job_skills`. Questions that count, rank or compare across many people or jobs go to `query_exports`, which runs a read-only query over the organisation's [exports](/output-integrations/exports), so its figures are as recent as the latest export.

Every tool needs the **TechWolf Agent** role in the TechWolf Console. The last column names the data access a tool needs on top of that role; where it names several, any one of them is enough. **Employee Org Units** and **Jobs** then decide which employees and jobs an answer includes. The TechWolf Console admin sets all of these for each person in **User Management**, as described in **Give people access to the Analyst Agent and the MCP connector** and **Configure user access management**.

| Tool | What it reads | Data access |
| --- | --- | --- |
| `tenant_overview` | How many employees, jobs and job families TechWolf holds for the organisation | **Skill Supply**, **Skill Demand** or **Tasks on Jobs** |
| `list_employees` | Employees found by name or job title, up to 50 per call, to look up the employee ID the other tools need | **Skill Supply** |
| `get_employee` | One employee's skill profile, with details such as their job and organisational unit | **Skill Supply** |
| `resolve_skills` | Skills found by name or looked up by ID, or the organisation's whole Skill Taxonomy | **Skill Supply** or **Skill Demand** |
| `job_architecture` | Job families and their jobs: a search by name, one job's details, or the whole job architecture | **Skill Demand** or **Tasks on Jobs** |
| `job_skills` | The skills a job needs, for one job or up to 20 at once | **Skill Demand** |
| `match_employee` | How one employee's skills match one job: a match score, the job's skills they have and the ones they miss | **Skill Supply** |
| `list_companies_vacancies` | The organisation's companies, their open vacancies and the skills a vacancy asks for | **Skill Supply**, **Skill Demand** or **Tasks on Jobs** |
| `job_tasks` | A job's tasks, each with its AI impact level and the reason behind it, or a search across the organisation's tasks | **Tasks on Jobs** |
| `task_similarity` | How closely the tasks of two jobs, or of two job families, match, or the jobs or job families whose tasks are most similar to one | **Tasks on Jobs** |
| `org_units` | Organisational units and how they nest: the units under one unit, or the whole structure | **Skill Supply**, **Skill Demand** or **Tasks on Jobs** |
| `courses` | The course catalogue, and the skills each course teaches | **Skill Supply**, **Skill Demand** or **Tasks on Jobs** |
| `query_exports` | Counts, rankings and comparisons across employees, jobs, skills, tasks, matches and courses | **Skill Supply** or **Skill Demand**, plus the access for each table a query reads: **Skill Supply** for employees, **Skill Demand** for jobs, both for matches between them, **Tasks on Jobs** for tasks |
| `market_search` | Public job postings: statistics, such as the skills that postings for a job title in a country ask for most, or a few example postings | The **TechWolf Agent** role alone |

## Prerequisites

The tools come with the MCP connector, so they need what the MCP connector page lists. The task tools, `job_tasks` and `task_similarity`, answer from **Work Intelligence**, and `market_search` from **Market Intelligence**, which every organisation has.

## Availability

The tools are the same in every AI tool the connector supports. Admins find them here:

- **Claude:** under **Customize** > **Connectors** > **TechWolf** > **Tool permissions**, where a Claude owner can set them to **Always allow**. See [TechWolf MCP connector in Claude](/agent-integrations/mcp-connector/claude).
- **ChatGPT:** in the list of tools ChatGPT finds when an admin adds TechWolf. ChatGPT keeps that list, so an admin selects **Refresh** in TechWolf's app settings to pick up new tools. See [TechWolf MCP connector in ChatGPT](/agent-integrations/mcp-connector/chatgpt).
- **Microsoft Copilot Studio:** on TechWolf's tool page, opened from the agent's **Tools** page. See [TechWolf MCP connector in Microsoft Copilot Studio](/agent-integrations/mcp-connector/copilot-studio).
- **Glean:** in TechWolf's **Edit settings** under **Platform** > **Tools** in the **Admin console**, where a Glean admin turns each tool on for Glean Assistant, for agents or for both. See [TechWolf MCP connector in Glean](/agent-integrations/mcp-connector/glean).

## Related pages

- [MCP connector security](/agent-integrations/mcp-connector/security)
- [Skills Intelligence](/skills-intelligence)
- [Work Intelligence](/work-intelligence)
- [Market Intelligence](/market-intelligence)