get-md CLI
get-md turns HTML pages, PDF, DOCX and Markdown into clean Markdown for language models, from the command line or code.
- 96 ★ on the repository
- Last commit
Extended access
Why this level
- The command fetches URLs and downloads pages and images to disk.
- The optional model mode sends page content to the configured model.
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/get-mdDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The command fetches URLs and downloads pages and images to disk.
- The optional model mode sends page content to the configured model.
This is third-party code. Review the repository files before installing.
What it does
get-md is a document-to-Markdown converter tuned for feeding a language model. It takes a file, a URL or stdin, detects the format (HTML, PDF, DOCX or Markdown) and returns structured text with a YAML metadata header. Pages are cleaned with Readability, and PDFs get their headings, paragraphs and lists rebuilt. The `getmd` command can strip images, links and tables, cap output length, download images to a folder and cache responses on disk. Mermaid blocks are preserved, and with syntax validation on, broken diagrams are flagged. There is also a library: `convertToMarkdown` in TypeScript. A local model can optionally be attached for conversion.
Who it is for. Developers of agents and RAG pipelines who need to feed a model clean text from docs, articles and PDFs.
Good fit when
- An agent has to read a docs page without menus, footer and ads.
- A PDF manual should become Markdown with headings and lists kept.
- A site with Mermaid diagrams moves into a knowledge base and diagram sources must survive.
- A batch of pages is converted by a script and repeat requests should come from cache.
Not a fit when
- Pages are built by JavaScript after load: the converter reads the fetched HTML.
- You need an MCP server for the agent: get-md exposes only a CLI and a library.
- The runtime is Node.js older than version 22.
Example request
Fetch https://example.com/docs with getmd, strip the images and save the result to docs.md.Limitations
Node.js 22 or newer is required. Scanned PDFs without a text layer give an empty result, because OCR is not claimed in the README. Rebuilding diagrams from a PDF goes through a remote vision model, is opt-in and approximate. Mermaid validation needs the separately installed `mermaid` package. Default output length is capped at one million characters. MIT license, package version 1.7.0 at the time of review. The sub-100 ms speed is claimed for the mode without a model.
How to disable. Running through `npx` leaves nothing in the system. Remove a global install with `npm uninstall -g @nanocollective/get-md`; the cache lives in `~/.get-md/cache`.
FAQ
How do I run get-md without installing?
Run `npx @nanocollective/get-md https://example.com -o output.md`. The result goes into the file; without `-o` it prints to stdout.
Which formats does get-md read?
HTML, PDF, DOCX and Markdown. The format is detected by file extension, URL or the `%PDF` signature on stdin.
How is it different from MarkItDown?
It focuses on feeding a model: Readability cleanup of pages, YAML metadata, Mermaid handling and an HTTP cache. It is a Node.js package with no Python needed.
How do I drop links and images from the output?
Add `--no-links` and `--no-images`. Tables are switched off with `--no-tables`.
Does get-md ship an MCP server?
The README does not claim an MCP server: a `getmd` command and a library are available. The agent calls the command from a shell.
Discussion
No questions or reviews yet.
Related
CLI toolsPython CLI and library that turn PDF, Word and PowerPoint into structured elements, JSONL pages and skills with no OCR or models.
A CLI engine that compiles documents into a local structured wiki: a transparent alternative to plain RAG, reachable by an agent over MCP
An MCP server gives agents search and answers over local documents using hybrid retrieval, reranking and section links.
Skill and CLI for tables: cleaning with recorded decisions, level-based statistics, a PDF report and a separate audit.
minirag-mcp searches your documents with hybrid vector and BM25 search and runs on your computer without sending files out
The official Graphlit MCP server: an agent collects material from Slack, GitHub, mail and websites into a knowledge base and searches it
README in short
A Nano Collective community project, the `@nanocollective/get-md` package under the MIT license. Conversion is claimed at under 100 ms without a model; with a model, quality rises at the cost of time.