fetchLABSCO SUMMARY
Where the content lands is your decision, not the tool's.
Two fetch tools that differ only in destination is a deliberate split: read a page when you need to reason over it, write it to disk when you only need it kept. The Markdown and cleaning modes cut the same page down further. The third tool is a plain HTTP client, which is what makes this usable against APIs rather than only against pages.
What it is
A local MCP server that makes HTTP requests for the model, with three tools: fetch a URL and read it back as Markdown or cleaned HTML, fetch the same URL straight to a file so it never enters the conversation, and issue an arbitrary request with your own method, headers and body.
What you get
- A page returned as Markdown, as raw HTML with the response headers, or cleaned to a level you choose
- The same fetch written to a file instead of the reply, for pages too large to read into context
- Any HTTP method — GET, POST, PUT, PATCH and DELETE — with query parameters, custom headers, and a text or JSON body
- The complete response back: status line, headers and body
- A User-Agent you set yourself, or one generated at random to constraints you give
Requirements
Python. No account, no key. File writes can be confined to the client's workspace root when the client supports that capability, and are otherwise done by absolute path.
Setup effort
Build from source — git clone https://github.com/coucya/mcp-server-requests.git
USE CASES
01
Reading an article without the markup
Fetch it as Markdown so the navigation, scripts and styling never reach the model.
02
Grabbing a page that would swamp the context
Send the fetch straight to a file, then read only the part you need from disk.
03
Calling an API that has no server of its own
Issue the request with your own method, headers and JSON body, and read the whole response back.
04
Checking what a server actually returned
Get the status line and headers alongside the body, rather than a rendered version of the page.
05
Getting past a site that varies by browser
Set the User-Agent yourself, or have one generated for a given browser and operating system.
3 tools
Listed in project docs. Your client may show a different list.
fetch_to_filehttp_requestBEFORE YOU INSTALL
What to know first
Tools you get
3 tools named in the project's documentation.
Requires
python on PATH
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/coucya/mcp-server-requests.git
cd mcp-server-requests
pip install .MCP Clients
Any MCP client
ANY MCP CLIENT · JSON
{
"mcpServers": {
"mcp-server-requests": {
"command": "python",
"args": [
"-m",
"mcp_server_requests"
]
}
}
}The block runs plain python -m mcp_server_requests, so "command" must be the same Python you ran pip install . with. If pip refuses to install into your system Python (an "externally-managed-environment" error), install into a virtual environment instead and put the full path of that environment's python in "command".
