config.toml¶
~/.config/genimg/config.toml holds your defaults and auth profiles.
| To | Run |
|---|---|
| Create it | genimg setup |
| Print it | genimg config show |
Open it in $EDITOR |
genimg config edit |
| Keep it elsewhere | export GENIMG_CONFIG_HOME=DIR |
default_model = "gdm:nb2" # used when --model is omitted
default_quality = "medium"
default_aspect_ratio = "16:9" # skipped if the model can't use it
default_resolution = "2K" # same
[profiles.google] # the name is yours: --profile google
provider = "google"
auth = "vertex_adc"
project = "my-gcp-project"
[profiles.work]
provider = "openai"
auth = "azure"
endpoint = "https://myres.openai.azure.com"
Keys and service-account paths stay in your environment, never in this file.
Auth modes¶
provider |
auth |
Route | Credentials | Settings |
|---|---|---|---|---|
openai |
direct |
OpenAI API | OPENAI_API_KEY |
|
openai |
azure |
Azure OpenAI | AZURE_OPENAI_API_KEY or OPENAI_API_KEY |
endpoint, api_version |
google |
direct |
Gemini API | GEMINI_API_KEY or GOOGLE_API_KEY |
|
google |
vertex |
Vertex AI | GOOGLE_APPLICATION_CREDENTIALS or CLAUDE_GCP_CRED |
project, region |
google |
vertex_adc |
Vertex AI | gcloud auth application-default login |
project, region |
codex |
subscription |
ChatGPT plan | codex login |
Without endpoint, genimg reads AZURE_OPENAI_ENDPOINT. Without project, it reads
GOOGLE_CLOUD_PROJECT, then the service-account JSON's project_id (vertex) or gcloud's active
project (vertex_adc).
Which profile runs¶
genimg authshows the winner in itssourcecolumn.
The first match wins:
--profile NAME--auth MODE(OpenAI), reusing that mode's profile settings- The provider's first profile
- Environment auto-detection