---
template: "page.peb"
title: "Analyze Image Skill"
description: "Use AI Image Understanding to analyze Salesforce image files with prompt-directed OCR, classification, description, and extraction."
displayName: "Analyze Image Skill"
category: "skills"
contentType: "reference"
audience: "admin"
tags: "skills,analyze-image,image-understanding,ocr,contentversion,agents,salesforce"
section: "skills"
seoTitle: "Analyze Image Skill"
seoDescription: "Analyze Salesforce images automatically with DialogueOne agents, including text extraction, visual questions, and HEIC conversion."
---

# Analyze Image

**Analyze Image** examines one Salesforce image version. Agents automatically receive this capability for supported
image files in Run Agent and file chat. You do not need to add it in Agent Studio. An existing configured Analyze
Image skill is reused when available. The prompt can request text, classification, visual description or field
extraction. The result can be Markdown, prose, tables or prompt-requested JSON.

## Supported files and routing

Analyze Image supports PNG, JPEG/JPG, WebP, nonanimated GIF, and HEIC/HEIF files up to 50 MiB. DialogueOne converts
HEIC/HEIF to JPEG when the agent needs to analyze the image. If a supported image exceeds a known model limit,
DialogueOne prepares a smaller copy where possible. Later prompts can reuse the copy for the same file version and
resize settings. The original Salesforce File stays intact.

The same preparation applies to file chat and to background runs with **Process File Before Running Agent** selected.
If preparation or analysis fails, follow the action in the failure message. Give your administrator the support
reference if the problem continues. PDFs and other documents follow their existing processing path.

## Inputs

| Input | Admin guidance |
| --- | --- |
| `contentVersionId` | Required exact Salesforce image version. |
| `contentDocumentId` | Optional matching Salesforce File ID. |
| `prompt` | Optional task-specific instruction; otherwise general analysis. |
| `detail` | `auto` by default; `low`, `high`, or `original` when needed. |
| `background` | `false` by default; controls the nested OpenAI request only. |

The running Agent supplies its configured model, reasoning, verbosity, and output limit. End users cannot override those
boundaries through this skill.

## Transactional and background use

Keep `background=false` for normal use, even when the agent runs in Background context. Automatic image handling uses
this setting to obtain completed findings. With `background=true`, a queued or in-progress result is only a submission
receipt. It does not mean that analysis finished.

OpenAI documents that Responses background mode temporarily stores response data to disk for roughly ten minutes even
when `store=false`. Do not use nested `background=true` where Zero Data Retention compatibility is required. Review the
[OpenAI data-controls guidance](https://developers.openai.com/api/docs/guides/your-data#background-mode) for the current
policy and image-input exceptions.

## Prompt and detail guidance

- For OCR: “Extract all visible text in readable Markdown. Preserve tables and report uncertain text.”
- For classification: name the categories and ask for visible evidence and uncertainty.
- For field or aviation photos: request objects, parts, condition, markings, damage, and relevant context.
- Use `low` for broad inspection and `original` for dense scans, small text, or spatially sensitive work.

## Cache, records, and cost

Completed results are cached by source, prompt, detail, effective model, and material Agent settings. Identical calls may
reuse the result. A new visual question can require a different analysis of the same image. For example, extracting
console letters does not identify all red indicators. Reformatting existing findings can use the conversation without
another analysis. Transcripts record the file version, settings, result and cache status.

Each call costs **5 DialogueOne skill credits**. A provider cache miss also incurs the selected model's image-input and
output-token charges; an DialogueOne result-cache hit avoids a new model call but still incurs the skill credits.

Analyze Image reads the image. Configure `dml_sobjects` or other business tools separately when the agent must create
or update Salesforce records. The agent must obtain the required findings before related record changes.
