[Developers](https://www.cv-transformer.com/sv-SE/docs/developers) [MCP server](https://www.cv-transformer.com/sv-SE/docs/developers/mcp-server)

# MCP server

2 min läsning

Connect AI assistants like Claude, ChatGPT and Cursor to CV-Transformer. They can then read and manage your candidates, layouts, team and webhooks, and turn CVs into candidates, through the Model Context Protocol (MCP).

## [Server address](#server-address)

`https://www.cv-transformer.com/mcp`

The server uses the Streamable HTTP transport. Its details are at `https://www.cv-transformer.com/.well-known/mcp/server-card.json`.

## [Connect with your account](#connect-with-your-account)

Most assistants connect with OAuth: add the server address, and the assistant sends you to CV-Transformer to sign in.

1. Add `https://www.cv-transformer.com/mcp` as a custom connector or MCP server in your assistant. In Claude, go to **Settings** → **Connectors** → **Add custom connector**.
2. Sign in to CV-Transformer when asked, with two-factor authentication if you use it.
3. Choose the organizations the assistant may use and what it may do there: read, or read and change, per area. You can only choose organizations where you manage API keys.
4. Click **Allow access**. You return to your assistant.

You can see and disconnect your apps under **Settings** → **Connected apps**. Members who manage an organization's API keys see the apps that have access to it on its API keys page, and can remove them there. An app loses access to an organization as soon as you lose the permission to manage its API keys.

## [Connect with an API key](#connect-with-an-api-key)

For scripts and clients without OAuth, send an API key as a bearer token:

```
Authorization: Bearer YOUR_API_KEY
```

Create a key under **Settings** → **API keys**. A key limited to scopes limits the tools in the same way.

## [Tools](#tools)

- **list_organizations**: the organizations the connection reaches.
- **describe_tables**: the columns and relations of tables, to read before querying them.
- **query_rows**, **insert_rows**, **update_rows**, **delete_rows**: read and change candidates, layouts, members, webhooks and organization settings with PostgREST queries, like the REST API.
- **create_candidate**: create a candidate. This uses one credit.
- **attach_candidate_file**: add a file, like a CV, to a candidate.
- **fill_candidate**: fill or adjust the candidate's layout with AI.
- **extract_candidate_data**: read structured data from a candidate's files, following a JSON schema.
- **invite_member**, **revoke_member**: manage your team.

Each tool needs the same scope as the REST API endpoint behind it, and requests count toward the same rate limit. Changes made through an assistant show in the activity log as the assistant acting for you.

## [Example prompts](#example-prompts)

- "Create a candidate from this CV and fill the layout."
- "Which candidates were added this week, and who added them?"
- "Extract the years of experience and languages from candidate Jane Doe."
- "Invite sam@example.com to the team, with permission to edit candidates."

See the [API reference](https://www.cv-transformer.com/en/api-reference) for the resources and scopes.

[FöregåendeWebhooks](https://www.cv-transformer.com/sv-SE/docs/developers/webhooks) [NästaConnected apps and OAuth](https://www.cv-transformer.com/sv-SE/docs/developers/connected-apps)