Humsana

Provides a local MCP server that integrates with the daemon to block dangerous commands and AI rewrites based on fatigue and safety rules.

typescript

0

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

{
  "mcpServers": {
    "sriramnatrajhen-humsana-mcp": {
      "command": "node",
      "args": [
        "/path/to/humsana-mcp/dist/index.js"
      ],
      "env": {
        "webhook_url": "https://hooks.slack.com/services/XXX/YYY/ZZZ"
      }
    }
  }
}

You protect your development workflow with an MCP server that safely routes and moderates commands and AI rewrites. This setup lets you run trusted, automated checks before executing potentially dangerous actions, helping you stay productive while avoiding risky changes.

How to use

To use the MCP server, you first run the local daemon and then run the MCP server locally. Your MCP client (such as Claude Desktop or another compatible client) should be configured to contact the MCP server so it can perform safety interlocks on commands and AI-generated file edits. You will typically operate through the client, which sends requests to the MCP server to evaluate commands or AI rewrites before they proceed.

How to install

Prerequisites: ensure you have Python and pip installed for the daemon, and Node.js and npm installed for the MCP server.

# Install the daemon that enforces safety interlocks
pip install humsana-daemon

# Start the daemon and keep it running in a terminal tab
humsana start

How to install (continued)

Install the MCP server globally so your client can access it.

npm install -g @humsana/mcp-server

Available tools

get_user_state

Query current fatigue, focus, and stress levels from the daemon to adjust safety checks.

check_dangerous_command

Assess whether a given shell command would be blocked under current safety rules.

safe_execute_command

Execute shell commands with an interlock that can block dangerous actions based on fatigue and thresholds.

safe_write_file

Apply AI-generated file edits with protection thresholds to prevent excessive or risky rewrites.