How to Configure Etendo Copilot to Use a Proxy

Overview

Etendo Copilot can be configured to route its LLM API traffic through a proxy for several operational reasons:

Network isolation (servers without direct internet access)
Centralized routing or load balancing, compliance and traffic inspection.
To consolidate provider credentials behind a controlled endpoint.

A proxy in front of the LLM providers acts as an intermediary that forwards requests from Copilot to providers such as OpenAI, Anthropic or Google.

This guide shows a common setup using LiteLLM Proxy and explains how to point Etendo Copilot to that proxy. The no direct internet access case is one typical scenario, but the same configuration applies when you need centralized routing, policy enforcement, or better observability of LLM requests.

Set up the proxy server

A convenient option is LiteLLM Proxy, which simplifies handling different LLM providers behind a single endpoint. The following example shows a basic Docker setup:

Create the litellm_config.yaml file: This file tells the proxy which models and provider behavior to expose. For a simple passthrough, use:

model_list:
  # Works for ALL Providers and needs the default provider credentials in .env
  - model_name: "*"
    litellm_params:
      model: "*"

Create the .env file: Store provider API keys securely in this file. Replace placeholders with your real keys:
```
OPENAI_API_KEY=sk-proj-XXXXXXX
ANTHROPIC_API_KEY=sk-ant-XXXXXXXXXX
GOOGLE_API_KEY=
```
Run the proxy with Docker: Place litellm_config.yaml and .env in the same directory and run:
```
docker run \
    --env-file .env \
    -v $(pwd)/litellm_config.yaml:/app/config.yaml \
    -p 4000:4000 \
    ghcr.io/berriai/litellm:main-stable \
    --config /app/config.yaml --detailed_debug
```
This command starts a container that exposes the proxy on port 4000 of the host server.

Info

The example above shows a simple passthrough configuration. For advanced usage (caching, access control, rate limiting, etc.) consult the LiteLLM Proxy documentation

Configure Etendo Copilot

After the proxy is running, configure Etendo Copilot to send its LLM requests through the proxy URL.

Edit gradle.properties: Open the gradle.properties file at the root of your Etendo installation.
Add the COPILOT_PROXY_URL property: Add the following line, replacing my.server.with.the.proxy and <PORT> with your proxy host and port (for the example above the port is 4000):
```
COPILOT_PROXY_URL=http://my.server.with.the.proxy:4000
```

Apply the changes and restart the service

To make the configuration effective you must rebuild, deploy the configuration to Etendo and restart the relevant services. Perform the steps in the order below:

Restart the Copilot service: From the project root run:
```
./gradlew resources.up
```
This target refreshes and restarts the Copilot service in your development/deployment packaging. Use this after changing runtime properties so the Copilot application reloads the new property values.
Deploy configuration to Etendo: Run the setup and smartbuild tasks to apply and build the configuration for Etendo:
```
./gradlew setup smartbuild
```
These tasks prepare and deploy the updated configuration and resources into the Etendo runtime so the platform uses the new COPILOT_PROXY_URL at the application level.
Restart Tomcat: Restart your Tomcat application server so Etendo Classic picks up the new configuration. The exact restart method depends on your OS and Tomcat installation.

After completing these steps, Etendo Copilot will send LLM requests through the configured proxy. This is useful for network isolation, centralized routing or load balancing, policy enforcement, and auditing.

This work is licensed under CC BY-SA 2.5 ES by Futit Services S.L.