Provider setup

Supported configuration

The bundled TUI stores provider settings in:

~/.config/hrns/config.json

If the file does not exist, startup falls back to an interactive onboarding flow.

First-run onboarding

On the first launch, hrns prompts for:

provider name
provider API URL
provider API key
default model
whether to skip TLS verification

The entered provider is saved and becomes currentProvider.

Config shape

The current file format looks like this:

{
  "providers": {
    "openai": {
      "url": "https://api.openai.com/v1",
      "key": "your-api-key",
      "model": "your-model",
      "skipVerify": false
    }
  },
  "currentProvider": "openai"
}

For the active provider, tui.Run builds openai.NewClient(...) with:

url as the OpenAI-compatible base URL
key as the bearer token
skipVerify to control TLS verification
model as the startup model shown in the TUI

Typical setups

Default OpenAI-style base URL

Use onboarding and enter:

provider name: openai
provider API URL: https://api.openai.com/v1
provider API key: your API key
default model: a model your provider supports
skip TLS verify: n

Custom endpoint

Use onboarding and enter your endpoint URL and model.

Local endpoint with self-signed TLS

Set:

provider API URL: https://localhost:8443/v1
provider API key: your development key
skip TLS verify: y

Important runtime detail

Provider configuration and model selection are separate concerns. You can inspect the models exposed by the current provider with:

/models

You can change the saved model for the current provider at runtime:

/model <your-model>

You can also add more providers with:

/connect

To switch the active session to a saved provider:

/provider <name>

/connect saves the new provider and marks it as currentProvider in the config file, but it does not rebuild the in-memory client by itself. Use /provider <name> to switch immediately, or restart the TUI to start with the saved current provider. In hrns exec, -provider <name> uses that provider’s saved default model unless you also pass -model.

When embedding instead of using the TUI

If you use the packages directly, you do not need the TUI config file at all. You can build the client explicitly:

client := openai.NewClient(
    openai.WithBaseURL("https://your-provider.example/v1"),
    openai.WithAPIKey("your-api-key"),
)

That is usually the cleaner path for real applications.

Getting started

Building with hrns

Project development

Provider setup

Supported configuration

First-run onboarding

Config shape

Typical setups

Default OpenAI-style base URL

Custom endpoint

Local endpoint with self-signed TLS

Important runtime detail

When embedding instead of using the TUI

​Supported configuration

​First-run onboarding

​Config shape

​Typical setups

​Default OpenAI-style base URL

​Custom endpoint

​Local endpoint with self-signed TLS

​Important runtime detail

​When embedding instead of using the TUI

Supported configuration

First-run onboarding

Config shape

Typical setups

Default OpenAI-style base URL

Custom endpoint

Local endpoint with self-signed TLS

Important runtime detail

When embedding instead of using the TUI