The MCP service is a protocol layer on top of the Inyett Automation API. It uses the same authentication and product entitlements as the Automation API.
MCP server
Prerequisites
To connect to Inyett MCP, you need:- Active access to the Inyett Automation API.
- Either an Inyett API key or a valid Visma Connect access token.
- An MCP client that supports remote MCP over Streamable HTTP.
Authentication
Inyett MCP supports the same credentials as the Automation API:- API key through
X-Api-Key.
- API key
Send the API key in the A generic MCP configuration using an API key:Replace:
X-Api-Key HTTP header.-
YOUR_INYETT_MCP_KEYwith your API key. -
mcp.inyett.comwithmcp-test.inyett.comwhen connecting to the test environment.
Client configuration
MCP clients use different configuration formats. Use the following MCP server URL:Claude Desktop
Claude Desktop
Claude Desktop connects to remote MCP servers through Custom Connectors. It does not use a
type value such as http or streamable-http for this setup.- Open Settings in Claude Desktop.
- Open Connectors.
- Select Add custom connector.
- Enter the Inyett MCP URL:
- Complete the authentication flow provided by Claude Desktop.
Claude Desktop uses the connector’s remote authentication flow rather than a local
claude_desktop_config.json entry for this remote server. Do not add credentials to the MCP URL.LibreChat
LibreChat
OpenAI Responses API
OpenAI Responses API
Connection flow
An MCP client normally performs the following operations:1
initialize
Establishes the MCP connection.
2
tools/list
Retrieves the tools available to the credential.
3
tools/call
Invokes a selected tool.
Tools
MCP exposes the following tools:tools/list only shows the company data tools that the credential is entitled to use. whoami is available to authenticated callers.Company lookup parameters
Tools can use parameters describing the company to retrieve or search for, such as:- Organisation number.
- Company name.
- Country of the target company.
- Search parameters supported by the selected tool.
Error responses
Security recommendations
- Store API keys and access tokens in secure secret storage.
- Never add credentials to source control.
- Do not expose credentials in browser-side client code.
- Use production credentials only with the production endpoint.
- Use test credentials only with the test endpoint.
- Limit the tools made available to an AI agent when the client supports tool allowlists.