Integrations & MCP Directory
Browse by category: Lurus product MCP, Switch built-in servers, AI clients, SDKs, and protocol compatibility. Each card links straight to its integration docs.
This page is data-driven
Lurus Product MCP
Lurus's own products expose their capabilities as agent tools via MCP servers, callable from MCP clients like Claude / Codex.
The agent execution engine exposed as an MCP tool (kova mcp serve), letting Claude/Codex drive persistent agents directly.
Memory read/write as agent tools: memory_add / memory_search / memory_delete, with parameters aligned to the REST API.
Trace / Replay exposed to agents — debugging and zero-cost replay callable from MCP clients.
Switch Built-in MCP Servers
The Switch desktop app offers one-click installation of common MCP servers, delivered dynamically across CLIs by visible_to.
AI Clients
Just fill in the base_url and API Key to start chatting — no code required. See each client's integration guide for full steps.
Desktop multi-model client — fill in base_url + Key and go.
Open-source web / desktop chat UI with custom OpenAI-compatible endpoints.
Native iOS / macOS client with a custom API domain.
General config for any OpenAI-compatible client such as NextChat / ChatBox.
SDKs & Ecosystem Protocols
Compatible with standard SDKs and protocols — migrate existing code by changing a single base_url line.
Fully compatible with Python / Node.js / Go — just change base_url and api_key.
LumenCheckpointer drops in for SqliteSaver, connecting to Kova persistence.
Lumen exports traces via OTLP to Grafana Tempo / Datadog APM.
Unified identity federation with enterprise IdPs via standard OIDC / OAuth2 integration.