> ## Documentation Index > Fetch the complete documentation index at: https://help.memoryplugin.com/llms.txt > Use this file to discover all available pages before exploring further. # Image Memories > Save an image as a memory, get an AI description, and recall it through the API export const ArticleInfo = ({author, lastUpdated}) => { const authorAvatar = author === 'Alara' ? '/images/author-alara.jpg' : author === 'asad' ? '/images/author-alara.jpg' : null; const formatDate = dateInput => { if (!dateInput) return ''; if (typeof dateInput === 'string' && !dateInput.match(/^\d{4}-\d{2}-\d{2}/)) { return dateInput; } try { const date = new Date(dateInput); const now = new Date(); const diffTime = Math.abs(now - date); const diffDays = Math.floor(diffTime / (1000 * 60 * 60 * 24)); if (diffDays === 0) return 'today'; if (diffDays === 1) return '1 day ago'; if (diffDays < 7) return `${diffDays} days ago`; if (diffDays < 30) return `${Math.ceil(diffDays / 7)} week${Math.ceil(diffDays / 7) > 1 ? 's' : ''} ago`; if (diffDays < 365) return `${Math.ceil(diffDays / 30)} month${Math.ceil(diffDays / 30) > 1 ? 's' : ''} ago`; return `${Math.ceil(diffDays / 365)} year${Math.ceil(diffDays / 365) > 1 ? 's' : ''} ago`; } catch { return dateInput; } }; return
{authorAvatar ? {author} : author?.[0]?.toUpperCase()}
Written by {author}
Last updated {formatDate(lastUpdated)}
; }; Not everything you want your AI to remember is text. A screenshot of a receipt, a whiteboard photo, a diagram someone sent you. Image memories let you save the picture itself as a memory. MemoryPlugin writes a description of what's in it, so your AI, which works in text, can recall the image later by what it shows. It works a bit like describing a photo to someone over the phone. You hold up the picture, and MemoryPlugin writes down what's in it so the description travels with the memory. Where that breaks: your AI mostly recalls the written description, not the pixels, and the link to the actual image is short-lived, as covered below. ## Saving an image In the dashboard, the **Add Memory** input on your home page has an image button (tooltip **Upload image memory**). It opens the **Upload Image Memory** dialog: 1. Drag and drop an image, or click to browse. 2. Optionally add your own description in the text box. This gives the AI a head start on what matters about the image. 3. Click **Upload**. You will see an **Image memory saved** confirmation. Upload Image Memory dialog with a dropzone reading Drag and drop or click to browse, and the hint PNG, JPEG, WebP, GIF up to 10MB ### Supported formats * **PNG, JPEG, WebP, GIF** * Up to **10 MB** per image Anything else is rejected with "Unsupported format. Allowed: PNG, JPEG, WebP, GIF", and files over the limit with "Image too large. Maximum: 10MB". ## The AI description, and why it takes a moment Processing runs in the background. After you upload, MemoryPlugin kicks off a job that generates a description of the image and creates the vector embedding used for search. This is why you'll briefly see an **IMAGE** badge with "Generating description..." on the memory card. Once the job finishes, the description is attached and the memory is fully searchable. If you typed your own description at upload time, that is stored right away. The AI-generated description fills in once processing completes. ## How recall serves an image When a tool reads your memories through the API, an image memory comes back differently from a text one. Instead of plain text, it returns an object with three things: * `content_type: "image"` so the client knows it's an image * `image_url`, a signed link to the image file * `image_description`, the text description used for recall The `image_url` is a **signed link that expires after 4 hours**. That is deliberate: the link is meant to be used soon after it's handed out, not stored and reused indefinitely. If a client needs the image again later, it recalls the memory again and gets a fresh link. The `image_description` does not expire, so the AI can reason about what the image shows even without loading the file. ## Where image memories work, and where they don't * **Dashboard only for now.** Image upload lives in the dashboard's Add Memory input. Save your images there. * **Not in shared buckets yet.** Inside a bucket shared with you, the image button is disabled ("Image memories aren't supported in shared buckets yet"). Text memories are fine. See [Shared Buckets](/features/shared-buckets). ## Next steps Organize image and text memories into buckets Bring memories in from a file or take them with you