Daniel Rosehill danielrosehill

MCP Infrastructure Tools: Proxies, Gateways, Routers & Federation

Last updated: December 12, 2025

A curated list of tools for Model Context Protocol (MCP) federation, routing, aggregation, and access control.

🌐 Gateways

Conda Environment Strategy for AMD ROCm GPUs

Managing conda environments for AI/ML workloads on AMD GPUs can quickly consume 100GB+ of disk space. The PyTorch ROCm stack alone is ~18GB, and each new environment duplicates this.

The Problem

Creating environments independently:

# BAD: Each env duplicates the full PyTorch ROCm stack (~18GB each)
conda create -n project1 python=3.12

ROCm 6.4.4 + PyTorch 2.7.1 Docker Image - AMD RX 7700 XT/7800 XT Compatibility

Image Details

Property	Value
Image	`rocm/pytorch:rocm6.4.4_ubuntu24.04_py3.12_pytorch_release_2.7.1`
Size	69.6 GB
ROCm Version	6.4.4
PyTorch Version	2.7.1

Chatterbox TTS Generation Benchmark

Date: 2024-12-07
Model: ResembleAI/chatterbox

Test Results Comparison

Local (AMD ROCm)

| Metric | Value |

Docker Layering For PyTorch ROCm In Action

When building multiple AI/ML Docker images that all need PyTorch with ROCm (AMD GPU) support, naive approaches can waste tens of gigabytes of disk space. This guide shows how Docker's layer sharing works and how to verify your images are efficiently layered.

The Problem: Misleading Image Sizes

Running docker images shows seemingly massive duplication:

REPOSITORY          TAG       SIZE

Preventing ComfyUI Custom Nodes from Overwriting PyTorch ROCm

The Problem

When using ComfyUI with an AMD GPU and PyTorch ROCm, installing custom nodes via ComfyUI-Manager (or manually via pip) can silently replace your ROCm-enabled PyTorch with the default CUDA version.

This happens because:

Custom nodes specify torch as a dependency in their requirements.txt
pip resolves this to the default PyPI torch package (CUDA version)
Your ROCm torch gets overwritten without warning

Fix: Easy Effects "Deep Noise Remover Not Available" on Ubuntu

The Problem

When using Easy Effects installed via apt on Ubuntu, the Deep Noise Remover effect shows:

Deep Noise Remover Not Available The software required for the Deep Noise Remover effect, "DeepFilterNet", is not installed.

Why This Happens

Speed Up Linux Boot by Deferring Services Until After Graphical Desktop

A common cause of slow boot times on Linux desktops is services that block the boot process unnecessarily. Many services (Docker, Samba, databases, etc.) don't need to be running before you reach your desktop—they can start immediately after the graphical environment loads.

This guide explains how to defer services to start after graphical.target on systemd-based systems like Ubuntu with KDE Plasma.

The Problem

When you run systemd-analyze blame, you might see something like:

	# Fix Electron App Sandbox Crashes on Linux

	Electron apps on Linux often fail to launch with errors like:
	- `signal=TRAP`
	- Sandbox-related crashes
	- Apps that work with `--no-sandbox` but fail otherwise

	## The Problem

	Electron's sandbox requires either:

	[af9dac22-676b-48fc-9cea-45ef1d877efe]
	Description=IP Cam Viewer
	fullscreen=true
	fullscreenrule=2
	ignoregeometry=true
	ignoregeometryrule=2
	noborder=true
	noborderrule=2
	position=5760,480
	positionrule=2