Quickstart

1. Install the CLI

SciToolkit is distributed as a Python package. Install it with pip:

pip install scitoolkit

Requires Python 3.12+. The CLI manages isolated environments for each toolkit, so you don't need to worry about dependency conflicts.

2. Find a toolkit

Browse the registry to find a toolkit, or search from the command line:

scitoolkit search exoplanet

3. Install the toolkit

Install a toolkit by name. SciToolkit will download it, choose the right execution environment (venv, conda, or Docker), and install all dependencies in isolation:

scitoolkit install aster

The toolkit lives at ~/.scitoolkit/toolkits/aster/ along with its isolated environment. Other toolkits get their own directories — there are no shared dependencies.

4. Configure (if needed)

Some toolkits require configuration before they can be used — API keys, paths to local data, etc. If yours does, the install step will tell you. Run setup interactively:

scitoolkit setup aster

See Configuration & Setup for details.

5. Serve to your agent

Start an MCP server with all your installed toolkits:

scitoolkit serve

You'll see an interactive TUI where you can toggle individual toolkits on and off and watch tool calls in real time. Or, configure your agent to launch the server automatically:

scitoolkit configure claude-code

This adds an MCP server entry to your Claude Code config. Restart Claude Code, and the toolkit's tools become available.

Use in Python directly

If you're building your own agent and don't need MCP, you can load toolkits directly:

from scitoolkit import load_toolkit

aster = load_toolkit('aster')
tools = aster.get_tools()

# Use with your agent framework
agent = Agent(tools=tools)

Useful commands