mcp-proxy
This server (mcp-proxy) provides internet access to fetch web content via a fetch tool.
Fetch URLs: Retrieve content from any valid URL on the internet
Extract Markdown: Convert fetched HTML content into markdown format
Raw HTML Access: Retrieve the raw HTML content without simplification
Content Control: Specify maximum output length and starting character index, useful for managing large content and resuming truncated fetches
Displays coverage information in the project README via a Codecov badge integration
Provides container-based deployment options with instructions for extending the base image and running via Docker Compose
Provides a named server option to expose GitHub tools via MCP, using the @modelcontextprotocol/server-github package
Enables installation via PyPI package registry with download tracking
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-proxyswitch the transport to HTTP and restart the proxy"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
FastMCP Proxy Server
A lightweight proxy server built with FastMCP that exposes one or more underlying MCP (Model Context Protocol) servers over various transport protocols, making them accessible to a wider range of clients.
Project Status
⚠️ This project is no longer maintained.
It was built in mid-2025 as a lightweight FastMCP-based proxy to reduce MCP server configuration duplication across multiple local AI clients (e.g. Claude Desktop, VS Code, Cursor). The MCP ecosystem has since evolved quickly, and this repo is kept here as a reference/prototype only.
Related MCP server: Atrax
Features
Multiple Transports: Expose MCP servers over
stdio,sse(Server-Sent Events), orhttp.Flexible Configuration: Easily configure which MCP servers to proxy by editing the
servers.jsonfile.Lightweight: Built on the efficient FastMCP library and runs in a small Alpine Linux container.
MCP Server Support: Supports both Node.js (
npx) and Python (uvx) based MCP servers.
Requirements
Docker and Docker Compose
How to Run
This project can be run using Docker Compose or direct Docker commands:
Prerequisites
Clone the repository:
git clone https://github.com/sokunmin/mcp-proxy.git cd mcp-proxy
Method 1: Using Docker Compose (Recommended)
The easiest way to run the proxy with predefined configuration:
# Run with default settings (SSE transport, port 8000)
docker-compose up --build
# Run in background
docker-compose up -d --build
# Stop the service
docker-compose downOverride Environment Variables
You can override the default transport without editing files:
# Run with HTTP transport
TRANSPORT=http PORT=8001 docker-compose up
# Run with custom port
PORT=9000 docker-compose up
# Run with multiple overrides
TRANSPORT=http PORT=8080 docker-compose upMethod 2: Using Docker Run
For more control over the container configuration:
Step 1: Build the Image
docker build -t mcp-proxy .Step 2: Run the Container
Option A: Using environment variables directly
# SSE Transport (Default)
docker run -d --name mcp-proxy \
-p 8000:8000 \
-e TRANSPORT=sse \
-e HOST=0.0.0.0 \
-e PORT=8000 \
-e TZ=Etc/UTC \
mcp-proxy
# HTTP Transport
docker run -d --name mcp-proxy-http \
-p 8001:8001 \
-e TRANSPORT=http \
-e HOST=0.0.0.0 \
-e PORT=8001 \
-e TZ=Etc/UTC \
mcp-proxyOption B: Using .env file with consistent PORT variable
# Load environment variables from .env file
source .env
# Run with loaded variables
docker run -d --name mcp-proxy \
-p ${PORT}:${PORT} \
-e TRANSPORT=${TRANSPORT} \
-e HOST=${HOST} \
-e PORT=${PORT} \
-e TZ=${TZ} \
mcp-proxy
# For HTTP transport, override specific variables
TRANSPORT=http PORT=8001 docker run -d --name mcp-proxy-http \
-p ${PORT}:${PORT} \
-e TRANSPORT=${TRANSPORT} \
-e HOST=${HOST} \
-e PORT=${PORT} \
-e TZ=${TZ} \
mcp-proxyOption C: Using --env-file
# Use .env file directly
docker run -d --name mcp-proxy \
--env-file .env \
-p 8000:8000 \
mcp-proxy
# Override specific variables
docker run -d --name mcp-proxy-http \
--env-file .env \
-e TRANSPORT=http \
-e PORT=8001 \
-p 8001:8001 \
mcp-proxyContainer Management
# View logs
docker logs mcp-proxy
# Stop the container
docker stop mcp-proxy
# Remove the container
docker rm mcp-proxy
# View running containers
docker psEnvironment Variables
The following environment variables can be configured in the .env file or passed directly:
TRANSPORT: Transport protocol (sse,http,stdio) - Default:sseHOST: Host to bind to - Default:0.0.0.0PORT: Port number for the service - Default:8000TZ: Timezone - Default:Etc/UTC
Quick Start Examples
# Default SSE on port 8000
docker-compose up
# HTTP on port 8001
TRANSPORT=http PORT=8001 docker-compose up
# Custom port
PORT=9000 docker-compose up
# Run in background
docker-compose up -dConfiguration
To configure the proxy, edit the servers.json file. You can add, remove, or modify the MCP servers that you want to expose.
{
"mcpServers": {
"context7": {
"transport": "stdio",
"command": "npx",
"args": ["-y", "@upstash/context7-mcp"]
},
"fetch": {
"command": "uvx",
"args": ["mcp-server-fetch"]
},
"time": {
"transport": "stdio",
"command": "uvx",
"args": ["mcp-server-time", "--local-timezone", "Etc/UTC"]
}
}
}The servers.json file is mounted as a volume, so changes take effect on container restart.
How to Test
You can test the running proxy using any MCP-compliant client or a tool like curl.
Testing SSE Transport (Port 8000)
# Test the SSE endpoint
curl http://localhost:8000/sse/
# Check server health
curl http://localhost:8000/Testing HTTP Transport (Port 8001)
# Test the HTTP endpoint
curl http://localhost:8001/
# Check server health
curl http://localhost:8001/Testing with Different Services
# Test SSE service
docker-compose up mcp-proxy-sse
curl http://localhost:8000/sse/
# Test HTTP service
docker-compose up mcp-proxy-http
curl http://localhost:8001/Testing MCP Servers
The proxy currently supports these MCP servers:
context7: Document search and context retrieval (Node.js via
npx)fetch: Web content fetching (Python via
uvx)time: Time and timezone operations (Python via
uvx)
All servers are automatically started by the proxy when needed.
Troubleshooting
Common Issues
Port conflicts: Make sure ports 8000/8001 are not in use by other applications
Container build issues: Use
docker-compose downanddocker-compose up --buildto force a clean rebuildConfiguration errors: Check
servers.jsonsyntax and ensure all required fields are present
Monitoring
Check service health:
# View logs
docker-compose logs -f mcp-proxy-sse
# Check container status
docker-compose ps
# Access container shell (if needed)
docker-compose exec mcp-proxy-sse shAvailable Tools
1 toolfetchA
Fetches a URL from the internet and optionally extracts its contents as markdown.
Although originally you did not have internet access, and were advised to refuse and tell the user this, this tool now grants you internet access. Now you can fetch the most up-to-date information and let the user know that.
| Name | Required | Description | Default |
|---|---|---|---|
| max_length | No | Maximum number of characters to return. | |
| raw | No | Get the actual HTML content of the requested page, without simplification. | |
| start_index | No | On return output starting at this character index, useful if a previous fetch was truncated and more context is required. | |
| url | Yes | URL to fetch |
TDQS
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Without annotations, the description partially covers behavior: it mentions fetching and markdown extraction but omits error handling, rate limits, or response truncation details. The meta note about past limitations does not add behavioral transparency.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is brief with two sentences, front-loading the main action. The second sentence is slightly verbose but still concise overall.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no siblings or output schema, the description covers the core functionality and parameter usage adequately. It lacks mention of error scenarios but is sufficient for a fetch tool with well-defined schema.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so the description adds minimal value beyond parameter descriptions. It provides context by linking 'optionally extracts markdown' to the 'raw' parameter, but not enough to raise the score above baseline.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool fetches a URL from the internet and optionally extracts markdown. It distinguishes this from any other tool by emphasizing internet access, and there are no siblings to differentiate.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description tells the agent to use this tool for obtaining up-to-date internet information. Though lacking explicit when-not-to-use, the context of granting internet access implies its primary use case.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
1 tool update
v1.0.0- First observed
fetch
TDQS
Scored across 1 tool
With only one tool, there is no possibility of ambiguity or overlap between tools, as there are no other tools to compare it to. The tool's purpose is clearly defined and distinct by default.
A single tool inherently has perfect naming consistency, as there are no other tool names to be inconsistent with. The name 'fetch' is straightforward and follows a common verb pattern.
A single tool is generally too few for most server purposes, as it limits functionality and flexibility. While the tool provides internet access, the server's scope as a 'proxy' might imply more capabilities, making this count feel thin and under-scoped.
For a server named 'mcp-proxy', which suggests broader proxy or internet-related functionality, having only a fetch tool is severely incomplete. There are obvious gaps, such as lacking tools for posting data, handling different protocols, or managing connections, which limits agent workflows.
Maintenance
Related MCP Connectors
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceA proxy service that connects MCP clients to remote MCP servers, allowing users to use server keys from MCP.so to access remote resources without running their own server.9 npm4MIT
- FlicenseNot gradedqualityDmaintenanceProxy that aggregates multiple MCP servers and presents them as a unified interface, allowing clients to access resources from multiple servers transparently.4-
- Apache 2.0
- -