Configuration

Set optional defaults and reusable prompt flags in Sengpt's local JSON configuration.

Find the configuration

Terminal
sengpt --config_file

The command prints the current JSON along with the config file and browser profile paths. Every setting is optional.

Example configuration

config.json
{
  "username": "You",
  "model": "",
  "default_mode": "interactive",
  "timeout_minutes": 45,
  "headless": true,
  "no_glow": false,
  "copy": false,
  "save": false,
  "delete": false,
  "preconfigured_prompts": {
    "readme": "Write a concise README for the supplied code.",
    "explain": "Explain the supplied material clearly."
  }
}

Reusable prompt flags

Keys inside preconfigured_prompts become command flags. Sengpt combines the configured text with piped input and any prompt supplied on the command line.

Terminal
cat app.py | sengpt --query --explain

Model selection

Leave model empty to use your account's default. Model names are ChatGPT web model slugs and may change over time. A command-line --model value overrides the configuration for that run.