Install judgevet

Status: draft.

Published judgevet 0.10.1 provides the library, CLI and judgevet-mcp command. The MCP runtime is optional.

Install the library

Use Python 3.12 or newer. In a virtual environment with pip, run:

python -m pip install judgevet

Alternatively, in an existing uv project, run:

uv add judgevet

The package includes judgevet/py.typed for type checkers. The base installation does not install the MCP runtime. See supported imports.

Run the CLI

With uv installed, run this from any directory:

uvx judgevet --help

The base library installation above also installs judgevet in that virtual environment. For a persistent standalone command with uv, run:

uv tool install judgevet

If judgevet is not on PATH, run uv tool update-shell and start a new shell. Confirm judgevet --help works before following the CLI task guides.

Install the optional MCP runtime

Follow Connect an MCP host for the optional extra, host configuration and tool discovery.

Run the published MCP command

The pinned launcher runs without a judgevet checkout.

Optional: run MCP from a source checkout

Use the development launcher.

Choose a host

Follow the recipe for VS Code, Cursor, Claude Code, Claude Desktop, or Codex. Pi uses CLI access. Host setup needs neither direnv nor a source checkout.

Configure Codex for a project

Follow the Codex recipe.

Verify the connection

Discover tools and make a call in your host.

When MCP does not connect

Use the safe startup/discovery checks.

For a first Python call, follow the first-judgment tutorial. For existing applications, use the sync or async recipe. Install/import failures are covered by troubleshooting.