tessera-web stores configuration in the settings table of the database (key TEXT PRIMARY KEY, value TEXT). Keys can be read and written via Settings in the web interface, or directly with tessera-web's settings API.
Key Default Description
ai.endpoint(empty) URL of the OpenAI-compatible VLM endpoint ai.model(empty) Model name for image/video/archive analysis ai.api_key(empty) API key (if required by the endpoint) ai.tag_prefixai/Prefix added to all AI-generated tags ai.max_tokens512Maximum tokens in the model's response ai.formatjsonResponse format (json or text) ai.subject(empty) Description of the collection, sent as context ai.prompt_image(built-in) Override the image analysis prompt ai.prompt_video(built-in) Override the video analysis prompt ai.prompt_archive(built-in) Override the archive analysis prompt ai.video_sheet_max_frames64Maximum frames per sprite sheet for video analysis ai.video_frame_selectionintervalFrame selection method: interval or scene
Key Default Description
embedding.text_endpoint(empty) URL of the text embedding endpoint embedding.text_modelnomic-ai/nomic-embed-text-v1.5Text embedding model name embedding.vision_endpoint(falls back to text_endpoint) URL of the vision embedding endpoint embedding.vision_modelnomic-ai/nomic-embed-vision-v1.5Vision embedding model name
Key Default Description
face.endpoint(empty) URL of the face detection endpoint face.model(empty) Face detection model name