readURLMarkdownLABSCO SUMMARY
One tool, and the content budget already thought about.
Passing raw HTML to a model wastes most of the context on markup; readability extraction plus Markdown conversion is the cheap fix, and this does exactly that and stops. The timeout and size cap are set in the server file, which is worth knowing if you plan to point it at long pages — a 1MB ceiling is generous for an article and tight for a document dump.
What it is
A Python MCP server that fetches an HTTP or HTTPS URL, extracts the main content with a readability algorithm, and converts it to Markdown.
What you get
readURLMarkdown(url)— one call from URL to clean Markdown- Main-content extraction, so navigation, sidebars and boilerplate do not reach the model
- Guarded fetches: a 30-second timeout and a 1MB cap on content length
- A declared user agent, so the sites you read can see what is asking
Requirements
Uv. Clone the repository — uv run installs the dependencies on first launch — then have the client run uv run --directory <directory> python read_url_mcp/mcp_server.py with PYTHONPATH set to the same directory.
Setup effort
Build from source — git clone https://github.com/takada-at/read_url_mcp
USE CASES
01
Reading an article in the conversation
Give the assistant the URL and let it work from the extracted Markdown rather than a paste.
02
Summarising documentation you found
Fetch the page directly so the summary is grounded in the current text.
03
Keeping a long page from flooding the context
Rely on the size cap and content extraction rather than pasting the whole page in.
1 tool
Listed in project docs. Your client may show a different list.
BEFORE YOU INSTALL
What to know first
Tools you get
1 tool named in the project's documentation. readURLMarkdown times out after 30 seconds and reads at most 1MB of content.
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
This one is built from source rather than installed with a single command.
Server setup
TERMINAL
git clone https://github.com/takada-at/read_url_mcp
cd read_url_mcpPoint your client at the checkout with the config below — <YOUR_PYTHONPATH> is the full path of the read_url_mcp folder the clone created.
MCP Clients
Any MCP client
ANY MCP CLIENT · JSON
{
"mcpServers": {
"read_url_mcp": {
"command": "uv",
"args": [
"run",
"--directory",
"<YOUR_PYTHONPATH>",
"python",
"read_url_mcp/mcp_server.py"
],
"env": {
"PYTHONPATH": "<YOUR_PYTHONPATH>"
}
}
}
}
