Use Kojen from the tools you already have
Every Kojen deployment is an OpenAI-compatible chat completions endpoint, so any SDK that talks to OpenAI can call the agents you train. And the whole dojo is an MCP server, so your coding assistant can train, evaluate and deploy them for you.
Call a deployed model
Base URL https://kojen.ai/api/v1/endpoints/<deployment-id>, a deployment key, and the SDK you use today. Streaming, tool calls and structured output pass through.
Vercel AI SDK
createOpenAICompatible with your deployment's base URL. Streaming and tools work.
LangChain
ChatOpenAI with base_url, in Python and JS. Plus Kojen's MCP tools for agents.
LlamaIndex
OpenAILike in Python, the OpenAI LLM in LlamaIndex.TS, both on your base URL.
OpenAI SDK
Change base_url and api_key in the official SDK. Nothing else changes.
Run the dojo from your assistant
MCP server https://kojen.ai/api/mcp. Sign in with OAuth (no key to copy) or send an API key.
Something else? Any OpenAI-compatible client or MCP client works the same way. The developer docs cover authentication, the REST API, rate limits and every tool.
Get an API key
Kojen is free to start, no card needed. Train and evaluate an agent, deploy it behind an OpenAI-compatible endpoint, and connect your tools over MCP.