What it does
Looks at an infographic image directly and transcribes every label, statistic, and callout in its actual reading order, since none of that content is otherwise visible to screen readers or search engines. Produces both a detailed alt text/long description and a plain-text summary block published near the image so the data is indexable and usable without seeing the graphic.
Install
This skill isn't packaged in an installable registry. It's a plain SKILL.md file you can copy directly. Copy it from the SKILL.md tab and save it at the path for your assistant:
Use it in your AI assistant
Claude Code
.claude/skills/infographic-text-extractor/ (or ~/.claude/skills/infographic-text-extractor/ for all projects)
GitHub Copilot (VS Code)
.github/skills/infographic-text-extractor/
OpenAI Codex
.codex/skills/infographic-text-extractor/
Cursor
.cursor/skills/infographic-text-extractor/ (or ~/.cursor/skills/infographic-text-extractor/ globally)
Compatibility
Requires the agent to view actual image content in detail, not just the filename or surrounding page text.
The complete SKILL.md for this skill:
---
name: infographic-text-extractor
description: Use when a user has an infographic, chart image, or data graphic and wants its text and data made accessible or readable to search engines.
---
# Infographic Text Extractor
## When to use
A post includes an infographic or data-heavy image whose text and figures are invisible to screen readers and search crawlers because they exist only as pixels.
## Procedure
1. Look closely at the infographic image and transcribe every piece of text: headings, labels, axis values, statistics, and callouts.
2. Preserve the reading order the design implies (top to bottom, left to right, or however the visual flow is structured), not just a jumbled list.
3. Write a detailed long-description or extended alt text that conveys the data relationships, not just the individual numbers in isolation.
4. Draft a shorter plain-text summary block to publish alongside the image, giving both search engines and sighted skimmers a text version of the same information.
5. Double-check transcribed numbers and labels against the image at least once before publishing, since a misread digit in a data graphic is a real error, not a stylistic one.
## Guardrails
- Never estimate or round a figure you can't clearly read in the image, flag it as illegible instead of guessing.
- Keep the long description factual and structured, not a marketing rewrite of what the data shows.