Connect your AI model

PrefixHarbor lets you pick the intelligence that powers your SQL formatting agent. You choose between a cloud API, a local model on your PC, or the built-in option. Each choice balances speed, privacy, and capability for your specific formatting needs.

Three ways to run your agent

Cloud AI provider

Use a standard API endpoint to access cloud models. Your prompt and file text travel to that provider, so keep sensitive data in mind. This option works well when you need to process long queries, but the text goes to the provider. You can test the connection to see response times and context window limits.

Local PC model

Run a model directly on your Windows machine. With a local model, prompts and file text stay on the PC; nothing leaves the machine.

Built-in model

Start with the built-in model that comes with PrefixHarbor. It requires no API keys to get started. This option is suitable for standard formatting tasks and edits. You can always switch to a cloud or local model later if your needs change.

Verify your connection

The Test button checks your chosen endpoint. It shows the response time and the available context window.

Match the model to the task

Use a bigger model for complex refactors or intricate logic. Send hard steps to a larger model using run_on_model. Smaller models work well for simple edits and quick formatting tasks. This flexibility lets you balance performance and cost effectively.

Quality varies between models, so always review the output. Use Ask first to pause before actions that change files. Check the diff to ensure the layout is correct. This approval mode gives you full control over every change.