SuperAPI AI Directory

alttext-mcp

🎖️ 📇 🏠 - Generate and translate image alt text and manage an AltText.ai image library.

Open website

About this tool

AltText.ai MCP Server is an MCP (Model Context Protocol) server designed to let AI assistants generate image alt text, manage the image library, and queue alt text generation for web page images through the AltText.ai API.

Purpose:

  • Generate and translate image alt text directly in AI assistant conversations.
  • Manage the AltText.ai image library, including searching, listing, viewing details, updating, and deleting images.
  • Perform accessibility audits on web pages and batch-generate alt text for images missing alt text.

Main capabilities:

  • Account management: view remaining credits, update the webhook URL, and more.
  • Generate alt text: support generating alt text for a specified image URL or local file, support specifying the language, and translate existing alt text.
  • Image library management: search, list, get details, update, and delete images.
  • Batch operations: process CSV files containing image URLs and batch-generate for web page images missing alt text.
  • Effects and processing: generation and translation consume account credits, and additional languages and image conversion may increase costs; generation can overwrite existing alt text; updates replace the provided fields; deletion removes images from the library.

How to use:

  1. Environment requirements: Node.js 22+ and an AltText.ai API key.
  2. Add the server in the MCP client configuration: use npx -y @alttext_ai/alttext-mcp.
  3. Supports Claude Desktop, Claude Code, Cursor, Windsurf, and any MCP-compatible client.
  4. Can be installed via the npm package @alttext_ai/alttext-mcp; the command-line tool alttext (@alttext_ai/cli) can also be used for account management and site API key operations.
  5. Provides the skills/alttext skill, which allows coding agents such as Claude Code and Codex to drive the alttext CLI, including safe operation guidance.
  6. Skills can be added via npx skills, or installed as a Claude Code plugin from the repository plugin marketplace.

Example prompts:

  • "How many credits do I have left?"
  • "Generate alt text for https://example.com/photo.jpg "
  • "Generate alt text in French and Spanish for this image"
  • "Search my images for 'product photo'"
  • "Generate alt text for images missing alt text on https://example.com "
  • "Process this CSV file containing image URLs"

Applicable scenarios:

  • Website accessibility optimization, batch-supplementing image alt text.
  • Multilingual content localization, translating image alt text.
  • Image asset management, searching, updating, and deleting image library content through an AI assistant.
  • Developer workflow integration, automating alt text generation through an MCP client or command-line tool.

Limitations and notes:

  • Generation and translation use account credits, and additional languages and image conversion may increase total consumption; it is recommended to check get_account before performing paid operations.
  • CSV import and page scraping are handled as background queues; acceptance of a request does not mean generation is complete, and the image library and configured completion notifications should be checked.
  • Tool annotations describe effects for client reference, but do not enforce confirmation.
  • This package uses stdio and runs on the machine that starts the MCP client; image and CSV paths point to that machine; the contents of selected files are uploaded to AltText.ai.
  • The npm scope is @alttext_ai (with an underscore), and the GitHub organization is alttextai (without an underscore); this is intentional, and both point to the same official package.
  • The package's cryptographic provenance can be verified via npm audit signatures, or a specific version can be pinned for reproducible installations.
  • Hosted deployments must rate-limit POST /register at a trusted edge using verified client addresses; the Node service deliberately ignores forwarded client address headers.
  • Tests use mocked fetch calls and require no API key or network access.