For the complete documentation index, see llms.txt. This page is also available as Markdown.
Integrations
Connect LLM providers and MCP servers to your project.
This screen connects AI integrations to your project: LLM providers (OpenAI, Anthropic, Ollama, Groq, Google, Mistral, OpenRouter, Grok) and MCP servers (Model Context Protocol) — so AI agents and the built-in AI chat can work with your backend.
Project → Integrations. LLM providers and MCP servers connected to the project.
What you can do here
The screen has two tabs, each its own list:
LLM — columns: Integration Name, Provider, Default Model, Enabled, Default. The Default badge is per environment (e.g. "Default for PROD"); the backend assigns it to the first integration you create. When several integrations are active and none is the default, the list shows a warning.
Add New Integration opens the provider gallery for the active tab. Clicking a row opens the integration's edit page, where you can also enable, disable, test, or delete it. Connection tests run automatically — the backend probes the provider and stores the result, so there is no manual confirmation step. Once an LLM integration is active, the project's AI features (like AI chat) can use that model.
