Using LiteLLM with Shift

DANGER

On March 24, 2026 at 10:52 UTC versions v1.82.7 and v1.82.8 of the litellm package on PyPI were found to be compromised with credential-stealing malware.

The recommended actions to take are:

• Remove/uninstall litellm 1.82.7/litellm 1.82.8 immediately.
• Check for litellm_init.pth in your site-packages/directory.
• Rotate ALL credentials that were present as environment variables or in config files on any system where litellm 1.82.8 was installed.

View more details and updates:

• https://futuresearch.ai/blog/litellm-pypi-supply-chain-attack/
• https://github.com/BerriAI/litellm/issues/24512

LiteLLM is an open-source proxy/gateway that provides a unified interface for accessing multiple LLM providers.

In this tutorial, you will learn how to configure LiteLLM, Caido, and Shift to use models from various providers that are not directly supported.

LiteLLM Configuration

The following Docker Compose file runs two services: LiteLLM and a PostgreSQL database for chat history persistence.

1. Save the following docker-compose.yml file and navigate to its directory:

services:
  litellm:
    build:
      context: .
      args:
        target: runtime
    image: ghcr.io/berriai/litellm:main-stable
    ports:
      - '4000:4000'
    environment:
      LITELLM_MASTER_KEY: sk-admin-key-1234567890
      DATABASE_URL: 'postgresql://llmproxy:dbpassword9090@db:5432/litellm'
      STORE_MODEL_IN_DB: 'True'
      STORE_PROMPTS_IN_SPEND_LOGS: 'True'
    depends_on:
      - db

  db:
    image: postgres:16
    restart: always
    container_name: litellm_db
    environment:
      POSTGRES_DB: litellm
      POSTGRES_USER: llmproxy
      POSTGRES_PASSWORD: dbpassword9090
    ports:
      - '5439:5432'
    volumes:
      - postgres_data:/var/lib/postgresql/data

volumes:
  postgres_data:
    name: litellm_postgres_data

2. With Docker running, enter the following terminal command:

docker-compose up

3. Navigate to http://localhost:4000/ui/login/ in your browser and login.

4. Select Models + Endpoints, Add a Model, configure your model provider details, and click on the Add Model button to save the configuration.

5. Next, select Virtual Keys, click on the + Create New Key button, configure the key, and click on the Create Key button to save the configuration.

Caido Configuration

1. Click on the account button in the top-right corner of the Caido user-interface, select Settings, and open the AI tab.
2. Add your virtual key to the OpenAI API Key field and the listening address of LiteLLM in the OpenAI Base URL (optional) field.

Shift Configuration

1. Next, navigate to the Models interface of Shift, select OpenAI from the drop-down menu, and click on the + Add Custom Model button.
2. In the pop-up window, select OpenAI from the drop-down menu, enter the alias of the model you created in LiteLLM, and provide the Model ID using the following syntax:

openai/<model>

3. Click on the Add button to save the configuration and then click on its sliding radio button to enable the model.

4. Once the model has been added, it will be available as an option in Shift's model selection drop-down menu.