HydraMCP

MCP server that routes prompts to cloud and local model providers from a single setup.
  • typescript

2

GitHub Stars

typescript

Language

4 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

HydraMCP connects Claude Code and similar clients to your model providers, consolidating cloud subscriptions and local models behind a single terminal interface. It lets you compare, consensus, and synthesize results across multiple backends without juggling keys or tabs across apps.

How to use

You interact with HydraMCP through your MCP client (for example Claude Code). Start HydraMCP your way, then connect via your client’s MCP setup. You can route prompts to specific backends, run parallel model comparisons, poll for consensus, or synthesize ideas from multiple models to produce a single, high-quality answer.

How to install

Prerequisites you need before starting: Node.js 18 or newer installed on your machine. An MCP client configuration ready to register your HydraMCP instance (for example Claude Code). A backend you want to use, such as a cloud provider proxy or a local model server.

Step 1: Set up a backend (choose at least one, ideally both for redundancy)

CLIProxyAPI (Cloud Models) enables your existing subscriptions to appear as a local API. You authenticate once per provider and HydraMCP can route requests to those backends.

Install and run CLIProxyAPI binaries and create a config that points to HydraMCP on localhost.

Ollama (Local Models) lets you run local models on your machine. Install Ollama and pull a model to use as a fast, quota-free backend.

Step 2: Install HydraMCP

git clone https://github.com/Pickle-Pixel/HydraMCP.git
cd HydraMCP
npm install
npm run build

Configuration and running tips

Create and edit your environment file to wire HydraMCP to your backends. The following shows typical values you would set in a local setup.

cp .env.example .env

# CLIProxyAPI (skip if not using)
CLIPROXYAPI_URL=http://localhost:8317
CLIPROXYAPI_KEY=sk-my-local-key

# Ollama (skip if not using)
OLLAMA_URL=http://localhost:11434

Register HydraMCP with Claude Code

In Claude Code, register HydraMCP so you can list your models and start sending requests through the MCP system.

claude mcp add hydramcp -s user -- node /path/to/HydraMCP/dist/index.js

Available tools

list_models

List all available models across configured backends to see your provider pool.

ask_model

Send a prompt to a single model and receive its response back for quick querying.

compare_models

Run the same prompt across multiple models in parallel and view results side by side.

consensus

Poll several models and have a judge model evaluate agreement, returning a single answer with a confidence score.

synthesize

Gather responses from multiple models and fuse them into a single, superior answer.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational