Setup

Connected in two minutes.

One base URL, one key, every tool. Pick your client below and follow the steps.

1

Create your API key

Sign in, open Dashboard → API Keys, create a key and copy it. Keys are shown once, so store them safely.

Open API Keys
2

Point Claude Code at LUCIF

Export the two environment variables in your shell (add them to ~/.zshrc or ~/.bashrc to persist):

bash
export ANTHROPIC_BASE_URL="https://lucif.ai"
export ANTHROPIC_AUTH_TOKEN="sk-your-lucif-key"
3

Run with the lucif-claude model

Use lucif-claude so the CLI accepts the model id; routing stays fully automatic behind it.

bash
claude --model lucif-claude

# or set it as the default for the session
claude config set model lucif-claude
Why -claude? Some CLIs validate that the model name contains "claude". lucif-claude is the identical routing core with a compatible name.
1

Open the developer config

In Claude Desktop: Settings → Developer → Edit Config (or the 3rd-party gateway panel on newer builds), then point the inference gateway at LUCIF.

2

Add the gateway configuration

json
{
  "inferenceGatewayBaseUrl": "https://lucif.ai",
  "inferenceGatewayApiKey":  "sk-your-lucif-key",
  "inferenceModels":        ["lucif-claude"]
}
Field names vary by version. The principle never changes: gateway base URL + your key + a model containing "claude".
3

Restart & select lucif-claude

Restart Claude Desktop, pick lucif-claude from the model picker, and ship.

1

Create opencode.json

Place it in your project root (or ~/.config/opencode/config.json for global use):

json
{
  "provider": {
    "lucif": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "LUCIF API",
      "options": {
        "baseURL": "https://lucif.ai/v1",
        "apiKey": "sk-your-lucif-key"
      },
      "models": {
        "lucif": { "name": "lucif" }
      }
    }
  },
  "model": "lucif/lucif"
}
2

Run opencode

bash
opencode
# then /models → select lucif
1

Open Kiro model settings

Kiro → Settings → AI Model → Add custom model. Choose an OpenAI-compatible provider and fill in:

settings
Base URL   : https://lucif.ai/v1
API Key    : sk-your-lucif-key
Model ID   : lucif-claude
2

Set as active model

Select lucif-claude in the Kiro chat/agent panel. Autopilot & powers specs now run on the lucif routing core.

1

Any OpenAI-compatible client

Set base_url and your key. It works with OpenAI SDKs, LangChain, LlamaIndex, cURL and more.

bash
curl https://lucif.ai/v1/chat/completions \
  -H "Authorization: Bearer sk-your-lucif-key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "lucif",
    "messages": [{ "role": "user", "content": "Hello, agent." }]
  }'
2

Python SDK example

python
from openai import OpenAI

client = OpenAI(
    base_url="https://lucif.ai/v1",
    api_key="sk-your-lucif-key",
)
res = client.chat.completions.create(
    model="lucif",
    messages=[{"role": "user", "content": "Hello, agent."}],
)
print(res.choices[0].message.content)