Using guidelines

Depicta maintains a curated and constantly improving library of prompt best practices — covering lighting, composition, style, diagram layouts, and more. The guidelines system is designed for AI assistants and LLMs to query programmatically, helping them write better prompts and choose appropriate models.

depicta uvx depicta npx depicta

How assistants use guidelines

AI assistants with the Depicta skill file search guidelines automatically before generating images. The workflow is:

1. Search — the assistant queries guidelines relevant to the task (e.g., “product photography”, “flowchart layout”).
2. Incorporate — the assistant weaves relevant best practices into the prompt it writes.
3. Generate — the improved prompt produces better results.

This happens transparently — you don’t need to do anything.

Search via CLI

depicta guidelines search "portrait lighting"

Returns matching guidelines with content and relevance scores.

Search via API

curl "https://api.depicta.ai/v1/guidelines/search?q=portrait+lighting" \
  -H "Authorization: Bearer dpct_YOUR_API_KEY"

{
  "guidelines": [
    {
      "content": "## Portrait lighting\n\nFor natural-looking portraits...",
      "relevance": 0.92
    }
  ]
}

What’s covered

Guidelines cover topics like:

• Photography styles and lighting techniques
• Illustration and art direction
• Icon and graphic design conventions
• Diagram layout and labeling
• Model selection — recommending appropriate tiers (economy, standard, premium) based on the type of visual being created
• Common pitfalls and how to avoid them

The library is continuously expanded and refined based on real-world usage patterns and model improvements.