Skip to main content

Installation

Requires Node.js 18 or later. Installs both the hubify CLI and the hubify mcp launcher (which starts the Hubify MCP server for AI coding assistants).

Homebrew (macOS)

curl (Linux / macOS)

This downloads the latest binary and installs it to /usr/local/bin/hubify.

Verify Installation

You should see a version number printed (e.g. 1.0.0). If you get command not found, ensure your npm global bin directory is in your PATH (see Troubleshooting below).

Update

Uninstall


Using with Claude Code

After installing the CLI, add the Hubify MCP server to Claude Code so Claude can read and write your lab data directly.
The hubify mcp launcher is available in hubify-labs 2.0.0 and later. Run hubify --version to confirm your installed version is current.

Option A: claude mcp add (one command)

Then set your lab context:
Find your CONVEX_URL and lab slug under Settings → Developer in the Hubify web app.

Option B: manual ~/.claude/mcp.json

Create or merge into ~/.claude/mcp.json:

Verify Claude Code sees the server

Restart Claude Code, then run:
Expected line: hubify connected 48 tools Run the health check to confirm env resolution:
For the full Claude Code setup walkthrough, see MCP Setup.

Using with Codex

After installing the CLI:
Or add to your Codex config file (~/.codex/config.json):

Requirements

Proxy / Firewall

If you are behind a corporate proxy:
The CLI communicates with www.hubify.com/api over HTTPS (port 443).

Troubleshooting

Ensure your npm global bin directory is in your PATH:
Avoid sudo npm install -g. Instead, configure npm to use a user directory:
Use nvm to install a newer version:
  1. Confirm hubify mcp --health passes all checks
  2. Confirm hubify --version works (CLI is on PATH)
  3. Check that CONVEX_URL and HUBIFY_LAB_SLUG are set in your mcp.json env block
  4. Restart Claude Code — MCP servers are loaded at session start, not hot-reloaded