Claude code

1. Claude Code + MaaS

Only supports Claude series models

Claude code

# Pull image    Node.js 18 or higher version
npm install -g @anthropic-ai/claude-code

# Environment variables
export ANTHROPIC_AUTH_TOKEN="{YOUR_API_KEY}"
export ANTHROPIC_BASE_URL="https://genaiapi.cloudsway.net/{YOUR_ENDPOINT}"
export ANTHROPIC_MODEL="claude-sonnet-4@20250514"

# Start claude code
claude    # Execute claude in any directory

2. Claude Code + LiteLLM + MaaS

Supports MaaS platform to access models in OpenAI format

1. postgres DB

docker pull postgres
docker run --name my-postgres -e POSTGRES_PASSWORD=123456 -p 5432:5432 -d postgres
docker exec -it my-postgres bash
psql -U postgres
CREATE DATABASE mydatabase;

2. LiteLLM

2.1 Configure config.yaml and place it in a custom folder (the path will be used in step 2)

model_list:
  - model_name: claude-3-7-sonnet-20250219 # Model name, parameter to be passed when calling the LLM interface
    litellm_params:
      model: claude-3-7-sonnet-20250219 # Actually called model name
      api_base: https://genaiapi.cloudsway.net/{endpoint}/v1/messages
      api_key: xxx # Automatically converted to x-api-key: Bearer xxx
  - model_name: gemini-2.5-flash
    litellm_params:
      model: gpt-3.5-turbo
      api_base: https://genaiapi.cloudsway.net/v1/ai/{endpoint}
      api_key: xxx
      headers: 
        Authorization: Bearer xxx # Authentication method for the api_base interface
general_settings: 
  master_key: sk-1234
  database_url: postgresql://postgres:123456@{db_url}:5432/my_database # Database configured in step 1

2.2 Start

docker pull ghcr.io/berriai/litellm:main-latest

docker run \
    -v $(pwd)/config.yaml:/app/config.yaml \
    -p 4000:4000 \
    ghcr.io/berriai/litellm:main-latest \
    --config /app/config.yaml --detailed_debug

2.3 Model Calling

# Check which models are configured in config.yaml
curl http://127.0.0.1:4000/v1/models

# Generate an authentication key for the model gemini-2.5-flash
curl 'http://0.0.0.0:4000/key/generate' \
--header 'Authorization: Bearer sk-1234' \
--header 'Content-Type: application/json' \
--data-raw '{"models": ["gemini-2.5-flash"], "metadata": {"user": "ishaan@berri.ai"}}'
# Response: {"key": "sk-wnm6JQ-f1U2Sixpa8L-0xg"}

# Call the model interface
curl --location --request POST 'http://127.0.0.1:4000/chat/completions' \
--header 'Authorization: Bearer sk-wnm6JQ-f1U2Sixpa8L-0xg' \
--header 'Content-Type: application/json' \
--data-raw '{
    "model": "gemini-2.5-flash",
    "max_tokens": 2000,
    "stream": false,
    "messages": [
        {
            "role": "user",
            "content": "What is the weather like in San Francisco?"
        }
    ]
}'

3. Claude code

3.1 Usage of claude code

# Pull image    Node.js 18 or higher version
npm install -g @anthropic-ai/claude-code

# Environment variables
export ANTHROPIC_BASE_URL=http://127.0.0.1:4000    # IP and port of litellm
export ANTHROPIC_MODEL=gemini-2.5-flash            # Which model to call
export ANTHROPIC_AUTH_TOKEN=sk-1234                # Authentication key

# Start claude code
claude    # Execute claude in any directory

3.2 claude code ui

http://{claude code ip}:{claude code port}/ui/

Username：admin
Password：sk-1234