Integration

Replit Agent with Meridian

Meridian exposes an OpenAI-compatible HTTP endpoint. Replit Agent speaks that protocol natively — swap the base URL and API key, and every model call routes through Nimbus infrastructure.

  1. 1

    Open Replit project secrets

    In your Replit workspace, open the Secrets tool (lock icon in the left sidebar). Secrets are injected as environment variables at runtime — never hard-code credentials.
  2. 2

    Add MERIDIAN_API_KEY

    Create a secret named MERIDIAN_API_KEY with the value nim_live_***. Generate this key from your Nimbus Dashboard → API Keys.
  3. 3

    Add MERIDIAN_BASE_URL

    Create a second secret named MERIDIAN_BASE_URL with the value https://meridian.getnimbus.net/api/v1. This is the OpenAI-compatible gateway — same path structure, same request/response shapes.
  4. 4

    Use any OpenAI SDK — change two lines

    Every major OpenAI SDK (Python, Node, Go) accepts a custom base URL. Replace the defaults with your secrets:

    Python

    from openai import OpenAI
    import os
    
    client = OpenAI(
        api_key=os.environ["MERIDIAN_API_KEY"],
        base_url=os.environ["MERIDIAN_BASE_URL"],
    )
    
    response = client.chat.completions.create(
        model="meridian-1",
        messages=[{"role": "user", "content": "Hello"}],
    )

    Node.js

    import OpenAI from "openai";
    
    const client = new OpenAI({
      apiKey: process.env.MERIDIAN_API_KEY,
      baseURL: process.env.MERIDIAN_BASE_URL,
    });
    
    const response = await client.chat.completions.create({
      model: "meridian-1",
      messages: [{ role: "user", content: "Hello" }],
    });

    That's it. Two lines changed — api_key and base_url. The rest of your code, including streaming, function calling, and multi-turn conversations, works unchanged.

  5. 5

    Replit Agent works as a drop-in

    Replit Agent uses the OpenAI protocol internally. When you set these two environment variables, every agent interaction — code generation, reasoning, tool use — flows through Meridian with no agent-level code changes. Your Replit workspace becomes a Nimbus-managed inference endpoint.

Nimbus · Meridian · Replit Agent integration · getnimbus.net