THE IMAGE-TO-ANYTHING

Snap a photo. Get back data. Receipts → expense lines. Whiteboards → typed notes. Screenshots → action items. Handwriting → text. The vision feature most people don't use, packaged for 5 daily use cases.

THE PROBLEM

You take photos of things to "deal with later." Receipts, whiteboards from meetings, signs with phone numbers, your kid's homework, the dosage on the back of the medicine bottle. Later doesn't come. The photos pile up.

Claude reads images. The skill is knowing the use cases.

THE SKILL

5 daily use cases:

Receipt → expense entry (pair with Day 39)
Whiteboard photo → typed notes
Screenshot of email → action items
Handwritten note → typed text
Sign / label / package → text + action

Each: snap, paste, done. Under 30 seconds.

INSTALL

Standard. Vision is built into Claude — no special setup.

THE FULL SKILL FILE

---
name: image-to-anything
description: Processes images into structured data. Five core patterns: receipts → expense lines, whiteboards → typed notes, screenshots → action items, handwriting → text, signs/labels → text + action. Designed for 30-second-or-less use cases.
when_to_use: User pastes or attaches an image, says "read this image," "transcribe," or describes wanting to extract info from a photo.
---

# The Image-to-Anything

You process images into useful output. Specific. Action-oriented.

## Inputs
1. **The image** (paste, attach, or describe location)
2. **What to do with it** (extract, summarize, convert)

If user just attached without context: ask which use case.

## 5 PATTERNS

### 1. RECEIPT → EXPENSE LINE

Vendor: [Name] Amount: $[X] Date: [MM/DD/YYYY] Category: [Auto-categorized — see Day 39] Spreadsheet row: [Date] | [Vendor] | [Amount] | [Category]


### 2. WHITEBOARD → TYPED NOTES

Recreated as structured text:

[Section 1 from board]
Bullet
Bullet
[Section 2]
Bullet
Decisions made: [if any]
Action items: [if any]


### 3. SCREENSHOT OF EMAIL → ACTION ITEMS

From: [Sender] Subject: [Subject] Date: [Date if visible]

What they want: [1 sentence] Action items for you: [Specific tasks] Suggested reply: [Use Day 5's 3-Word Email Reply if quick yes/no] Deadline: [If mentioned]


### 4. HANDWRITING → TEXT

Transcription: [Faithful text of handwritten content]

Notes:

Confidence: [high / medium / low based on legibility]
Anything unclear: [list specific words]


### 5. SIGN / LABEL / PACKAGE → TEXT + ACTION

Text on image: [Verbatim] Type: [Sign / label / package / etc.] Suggested action:

If phone number: "Call [number]"
If address: "Save to contacts? Map?"
If medication: "Dosage: [X]. Schedule reminder?"
If product: "Add to grocery list?"


## What NOT to do

- Don't summarize when transcription is needed
- Don't fabricate text in low-confidence sections — flag as unclear
- Don't process images of people without context (privacy)
- Don't extract sensitive info from documents (full credit cards, SSNs) — flag and ask user to crop

## Privacy notes

Always include for any image with personal data:

🔒 PRIVACY: This image contains [personal info type]. Claude has read it for this conversation. The image isn't permanently stored beyond your session.


## Delivery

End with this line, exactly:

---
*Snap, paste, done. The phone-photo backlog gets shorter every time.*

SAFETY CHECK

Image data is processed during conversation. Be especially careful with:

IDs / passports / driver's licenses
Credit cards
Medical records
Anything containing other people's faces or info

WHAT'S NEXT

Day 39 of 100. Pair with Day 39 — Receipt-to-Tax-File (the most common image use case) and Day 38 — Phone Photo Organizer.