Unifyr
Unifyr

Developer guide

Connect your workspace

Use MCP for an AI client, or the partner API for your own application. Start with a Unifyr account, the correct workspace and connected services.

Get started Explore integrations

MCP / ChatGPT

https://unifyr.nl/api/mcp
  1. Add the server URL in a compatible MCP client and choose OAuth authentication.
  2. Sign in to Unifyr, select your workspace and authorize read access.
  3. Ask which services are connected. Connect any missing services in Unifyr before using their tools.
  4. For Street View, use unifyr_show_street_view to display the image inline. Image cards require a client that supports MCP Apps.

ChatGPT directory review is pending. Manual connection depends on the custom connector features available in your ChatGPT account.

Partner API

Create an API key with the required scopes in your workspace. Keep it on your server; never put it in frontend code. This read request lists the agents accessible to the key.

bash
curl https://unifyr.nl/api/partner/v1/agents \
  -H "Authorization: Bearer $UNIFYR_API_KEY"

The @unifyr/connector SDK supports application integration. Use the current connector version and the workspace configuration required by your host application.

Permissions and changes

MCP read access does not authorize writes. Sending or changing data requires mcp:write and the prepare/execute confirmation flow. Revoke a connection under External connections. If a write result is uncertain, check the destination before retrying.

Troubleshooting

Check the selected workspace, connection status, permissions and available balance. Refresh your client's tool list after an update. Responses and supported operations vary by provider; use the discovered tool schema. Contact support with the time and error message, without tokens or API keys.

Support →