About formatika
formatika is an MCP server for processing local media and document files through an agent. It offers conversion and editing tools across formats, with files returned to the machine after processing.
Key features
- 01Convert and resize images
- 02Compress, crop, and rotate files
- 03Process audio and video
- 04Work with PDFs and archives
- 05Handle subtitles and tables
- 06Choose output directories
Use cases
- 01Convert HEIC photos to JPEG
- 02Compress a folder of images
- 03Create a square avatar from a portrait
- 04Process media files from an agent
MCP support
The server exposes formatika's file-processing tools to MCP clients. Its tool list is fetched from the service at startup; tools accept local file paths and an optional output directory.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y @formatika/mcpFORMATIKA_API_KEYSecretAccess key from your formatika.app account. Without it the free daily allowance applies, which is enough for occasional work.FORMATIKA_URLBase URL of the service. Only needed when pointing the server at a self-hosted instance.Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project README@formatika/mcp
Convert, resize, compress, crop and rotate files straight from your agent, using formatika.app.
Your agent gets the tools; the files stay on your machine until a job needs them and come back as soon as it is done.
Use it
No installation and no account needed to start:
npx @formatika/mcp
Claude Code
claude mcp add formatika -- npx -y @formatika/mcp
Claude Desktop, Cursor and other clients
Add this to the MCP servers section of the client configuration:
{
"mcpServers": {
"formatika": {
"command": "npx",
"args": ["-y", "@formatika/mcp"]
}
}
}
On Windows, Claude Desktop launches the server without a shell and cannot find
npx directly. Use this instead:
{
"mcpServers": {
"formatika": {
"command": "cmd",
"args": ["/c", "npx", "-y", "@formatika/mcp"]
}
}
}
Fully restart the client after editing the configuration.
Then just ask
Compress every photo in ./screenshots and put the results in ./web
Convert all the HEIC files in this folder to JPEG
Make a square 512×512 avatar out of portrait.png
Tools
The tool list comes from the service at start-up, so it never lags behind what formatika can actually do — this README does not have to be updated for a new tool to appear in your agent.
Today that is images, audio, video, PDF, archives, subtitles and tables: over thirty tools. The current list, with what each one does, is at formatika.app/mcp.
Every tool takes paths — one or more files — and an optional outputDir.
Results are written next to the source file unless you say otherwise, and
existing files are never overwritten: a name that is taken gets a number.
Limits and keys
Without a key you get a daily allowance, which is enough for occasional work. For volume, create a key at formatika.app/account and pass it in:
{
"mcpServers": {
"formatika": {
"command": "npx",
"args": ["-y", "@formatika/mcp"],
"env": { "FORMATIKA_API_KEY": "fk_..." }
}
}
}
FORMATIKA_URL points the server somewhere else — useful if you run formatika
yourself.
Privacy
Files are processed and deleted: two hours without an account, twenty-four hours with one. They are not looked at, not shared and not used to train anything. See the privacy policy.
License
MIT
About this repository
This is a read-only mirror of the packages/mcp directory of the formatika
monorepo, published so that the package has a public home: catalogues link to
it, and people who read code before installing it can do so.
Pull requests here cannot be merged — the next sync would overwrite them. Issues, however, are read and welcome: bugs, missing tools, awkward APIs.
Mirrored from commit e06be38.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.