Skip to content

Connect in a few minutes

You need Python 3.14 or later, uv, and Lightcast client credentials. Request credentials through Lightcast API access.

Install this version

Version 0.4.1 is published on PyPI. Run it without a repository checkout:

uvx --python 3.14 mcp-lightcast@0.4.1 --version

Alternatively, install into an existing Python 3.14+ virtual environment with python -m pip install "mcp-lightcast==0.4.1", then use mcp-lightcast directly. Existing 0.2 users should read the migration guide before upgrading.

Create a local .env file and set:

LIGHTCAST_CLIENT_ID=your-client-id
LIGHTCAST_CLIENT_SECRET=your-client-secret

Use LIGHTCAST_OAUTH_SCOPE=lightcast_open_free if that is the scope assigned to your free-tier credentials. Existing emsi_open accounts can keep the default.

Check your connection

uvx --python 3.14 mcp-lightcast@0.4.1 --env-file .env --validate-config

This makes one authentication request using your default scope. It does not prove that your account can use every premium API. For a broader check:

uvx --python 3.14 --from mcp-lightcast==0.4.1 python -m mcp_lightcast.diagnostics

Connect an MCP client

Use the following configuration in a client that supports local MCP servers. Replace the environment-file path with your own absolute path. If the client cannot find uvx, use its absolute executable path.

{
  "mcpServers": {
    "lightcast": {
      "command": "uvx",
      "args": [
        "--python", "3.14", "mcp-lightcast@0.4.1",
        "--env-file", "/absolute/path/to/lightcast.env",
        "--transport", "stdio", "--quiet"
      ]
    }
  }
}

The default transport is stdio. Startup messages and logging use stderr so stdout remains available for MCP messages.

Core LMI

The same installation includes all five Core LMI tools. Start with list_core_lmi_datasets to discover your account's catalog, then follow the Core LMI examples. These tools request the agnitio scope automatically; access depends on your Lightcast license.

Try a first question

Search for Python in the Lightcast skills taxonomy. Show its ID and type, then find five related skills.

Your client should call search_skills, followed by get_related_skills with the returned ID. This keeps identifiers grounded in real results.

Prefer HTTP?

uvx --python 3.14 mcp-lightcast@0.4.1 --env-file .env \
  --transport streamable-http --host 127.0.0.1 --port 3000

Connect to http://127.0.0.1:3000/mcp. This server uses Lightcast OAuth for outbound requests; it does not add authentication for inbound MCP clients. Keep it on localhost or place it behind an authenticated gateway.

Since version 0.4.0, the server also supports optional caller-supplied HTTP credentials and custom providers for authorized secret stores. See credential providers for configuration and account isolation. Existing local installations can keep the default environment mode.

Prefer Docker?

The published container includes Python and starts with stdio:

docker run --rm -i --platform linux/amd64 --env-file .env \
  ghcr.io/lawwu/mcp-lightcast:0.4.1

The image targets Linux amd64; Docker can emulate it on Apple Silicon. See deployment for HTTP hosting.

Work from source

For development, clone the repository and install the locked dependencies:

git clone https://github.com/lawwu/mcp-lightcast.git
cd mcp-lightcast
uv sync --python 3.14 --all-groups --frozen
cp .env.example .env
# Add your credentials to .env before calling the API.
uv run mcp-lightcast --transport stdio