About markitai
markitai converts documents, images, and web pages into Markdown for developers and agents. A shared Rust core powers its CLI, local service, browser workspace, MCP server, and language bindings.
Key features
- 01Document and web page conversion
- 02Image conversion with OCR
- 03Single-item and batch processing
- 04CLI, local service, and browser workspace
- 05Node.js, Python, and Go bindings
Use cases
- 01Convert a folder of documents to Markdown
- 02Extract text from a scanned image
- 03Prepare web content for agent workflows
MCP support
The MCP server exposes single-item, URL, and batch conversion tools for MCP clients.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx markitaiMODELprovider/model for --llm enhancement, e.g. gemini/gemini-flash-lite-latest; with the matching API key env var setMARKITAI_HOMEMarkitai's configuration, cache and model directory (default ~/.markitai)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 READMEMarkitai
Markitai converts documents, web pages and images to Markdown. A single Rust
core powers the markitai command (alias mkai), a local REST service with a
browser workspace, an MCP server, and in-process Node.js, Python and Go
bindings. The CLI executable needs no Python, Node.js or Go runtime; language
bindings need their respective language runtime.
Markitai 2.0 is the Rust rewrite of Python markitai 1.2. See the changelog and compatibility contracts.
Install
macOS and Linux:
curl -fsSL https://markitai.dev/setup.sh | sh
Windows (PowerShell):
powershell -ExecutionPolicy ByPass -c "irm https://markitai.dev/setup.ps1 | iex"
The setup scripts download the release archive for your machine from
GitHub Releases, check it
against the release's SHA256SUMS, and install markitai, mkai and
markitai-mcp for the current user, without administrator rights. Releases
cover macOS 11+ (Apple silicon, Intel), Linux with glibc 2.28+ (x86-64,
ARM64) and Windows (x64, ARM64). Python users can instead run
pip install markitai or uv tool install markitai (Python 3.10+). PyPI has
wheels for these six platforms only and no 1.x any more: elsewhere pip has
nothing to install and leaves an installed 1.x as it is. Manual downloads,
updates, uninstalling and going back to 1.x are in the
installation steps.
The archives include license notices and selected Markdown guides for offline use. OCR models, Chromium and LibreOffice are optional separate components. See platform support for requirements, unsigned macOS builds for Gatekeeper, and bindings for language packages.
Quick start
markitai report.docx -o out/ # writes out/report.docx.md
markitai ./documents -o out/ # whole folder, with a report
markitai scan.png --ocr -o out/ # local OCR; models may need preparation
markitai report.docx -o out/ --json # structured result for automation
markitai doctor # inspect optional components
markitai --help # commands, options and examples
markitai serve # browser workspace on 127.0.0.1:3600
For a separate trial state, set MARKITAI_HOME before using the CLI:
export MARKITAI_HOME="$PWD/.local/try-home". Plain document conversion does
not need a model connection. Existing configuration still applies; use
--no-llm -b native for local document processing without model enhancement.
URL inputs still need the network, even with --no-remote-fetch. Local OCR uses the portable Paddle engine on every
platform (macOS Vision with MARKITAI_OCR_BACKEND=vision); doctor --fix can prepare missing Paddle
models before offline use. JavaScript pages need Chrome/Chromium; Office page
images need LibreOffice. Model enhancement (--llm) uses configured provider
credentials or a subscription runtime.
Documentation
- Quick start: installation and the first ten tasks.
- CLI and MCP: command and Agent contracts.
- Agent index and standalone Agent guide: plain Markdown entry points, also included in CLI archives.
- Documentation index: the full source-checkout documentation. The offline archive contains selected guides; other topic links require this repository checkout.
License
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.