home / mcp / promptheus mcp server

Promptheus MCP Server

Exposes prompt refinement, tweaking, and provider/model inspection as MCP endpoints for integration.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "abhichandra21-promptheus": {
      "command": "promptheus",
      "args": [
        "mcp"
      ]
    }
  }
}

You will deploy and operate the Promptheus MCP server to expose intelligent prompt refinement capabilities through standardized MCP endpoints. This server lets clients refine prompts, tweak prompts, and inspect models and providers in a consistent way, enabling seamless integration with automation, UIs, and workflows.

How to use

Run the MCP server from your environment to expose prompt refinement capabilities to MCP clients. You can start a remote HTTP endpoint or use a local standard I/O interface, depending on your deployment and integration needs. Clients can request prompt refinement, apply targeted tweaks, and discover available models and providers through the MCP tools exposed by the server.

How to install

Prerequisites you need before installation:

Python 3.8+ is required
pip is available in your Python environment

Install the Promptheus MCP-enabled package and its runtime dependencies:

pip install promptheus
```

```bash
# Install the MCP runtime support if you plan to run the MCP server directly
pip install mcp

Starting the MCP server

You can run the MCP server in two common ways. Choose the one that best fits your workflow.

# Start the MCP server via the Promptheus CLI
promptheus mcp

# Or start the MCP server directly with Python
python -m promptheus.mcp_server

Available tools

refine_prompt

Intelligent prompt refinement with optional clarification questions. Accepts a prompt and optional answers to guide the refinement and returns a refined prompt with an action hint.

tweak_prompt

Apply targeted modifications to an existing prompt. Provide the current prompt and a description of changes to produce a modified version.

list_models

Discover available models from configured providers. Returns the availability and model list per provider.

list_providers

Check provider configuration status and the currently selected model for each provider.

validate_environment

Test environment configuration and API connectivity for configured providers.