---
name: amt-image-to-prompt
description: Turn a reference image into a structured generation prompt package with Agent Media Tools. Use when an agent needs visible-detail analysis, positive and negative prompts, and model-oriented settings without claiming to recover the original prompt.
---

# Image to prompt

Use `POST https://agentmediatools.com/api/prompt-director/from-image` with multipart field `image` (max 8 MB) or JSON `url`/`image_url`. MCP agents can call `image_to_prompt` with a public `image_url` and optional direction, medium, framing, aspect ratio, subject count, target model, interface, and advanced optimizer settings.

This is a paid vision operation: the standard call uses 2 flexible credits; the optional advanced optimizer adds 1 credit only on success. Ask before enabling `advanced` when cost intent is unclear.

Describe visible content only. Do not claim the output is the source image's original prompt, preserves identity, or guarantees reproduction. Return positive and negative prompts, recommended settings, warnings, and the requested target interface. Use `describe_image` instead when the user only needs a caption.

Tool: https://agentmediatools.com/tool/image-to-prompt
Guide: https://agentmediatools.com/guides/claude-image-to-prompt-skill