fake-py-mcp 0.3

Purpose

It’s easy for a model to look smart “on paper”, but real-world work requires following strict technical rules.

This MCP server provides a wide range of complex, “fake” tools to test how well a model actually follows instructions. By asking a model to generate data for these tools, you can see if it:

1. Uses the exact parameters required (no more, no less).

2. Provides the correct data types and formats.

3. Understands complex technical requirements without making mistakes.

If a model can accurately handle these fake tools, you can trust it to handle your real production tools. It’s a simple way to find out which models are actually ready for work.

Features

• 80+ fake.py fake data methods (names, text, internet, files, primitives, dates, geo, books, banking, file content), dynamically registered (no manual boilerplate), categorised and documented.

• Binary outputs (e.g., images, PDFs) are returned as base64-encoded strings.

• server_info tool for introspection.

Prerequisites

Python 3.10+

Installation

uv tool install fake-py-mcp

Usage

brew install mcp-inspector

CLIENT_PORT=8006 mcp-inspector fake-py-mcp

DANGEROUSLY_OMIT_AUTH=true CLIENT_PORT=8006 mcp-inspector fake-py-mcp

uv tool install mcpo

mcpo --port 8006 -- fake-py-mcp

Documentation

• Documentation is available on Read the Docs.

Tests

Run the tests:

pytest

Writing documentation

Keep the following hierarchy.

=====
title
=====

header
======

sub-header
----------

sub-sub-header
~~~~~~~~~~~~~~

sub-sub-sub-header
^^^^^^^^^^^^^^^^^^

sub-sub-sub-sub-header
++++++++++++++++++++++

sub-sub-sub-sub-sub-header
**************************

License

MIT

Support

For security issues contact me at the e-mail given in the Author section.

For overall issues, go to GitHub.

Author

Artur Barseghyan <artur.barseghyan@gmail.com>