Skip to main content

GitHub Copilot CLI

You can configure the GitHub Copilot CLI to use local GoingMerry model endpoints. This allows you to generate shell commands, git commands, and general terminal explanations entirely locally and offline.

1. Local Model Setup​

Ensure you have a suitable code-instruct or reasoning model running locally:

merry run gemma4

2. Configuration Mappings​

Copilot CLI commands use a local configuration file to route completions. Set the host endpoint inside your environment configuration or custom client wrapper:

export COPILOT_CLI_API_URL="http://localhost:11434/v1"

Because GoingMerry provides a standard OpenAI-compatible API at /v1, the Copilot CLI will automatically route request payloads to the local daemon.

3. Explaining Commands Locally​

Once mapped, you can explain commands in your shell:

github-copilot-cli explain "tar -xvf archive.tar.gz"

The prompt is resolved locally by the active gemma4 model with zero latency and high data privacy.