OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Frameworks

Tooluniverse

An open-source Python framework for building AI scientist systems and connecting language models to scientific tools, datasets, APIs, packages and MCP servers.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Tooluniverse — OpenViST project profile in Agent Frameworks.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when an AI-agent project needs a programmable scientific tool layer, MCP integration or composed research workflows rather than a standalone end-user application.

Who it is for

  • Developers building AI scientist agents.
  • Research teams prototyping automated scientific investigation.
  • Python developers integrating scientific tools into agent workflows.
  • Projects that need an MCP-compatible scientific tool layer.

Documented uses

  • Connect language models to domain-specific scientific tools.
  • Create AI co-scientist or autonomous research prototypes.
  • Expose scientific utilities through MCP servers.
  • Discover, inspect, run and test tools from a terminal.
  • Compose tools into sequential or parallel workflows.
Before you choose

Requirements & limits

The repository is public, active and licensed under Apache-2.0. The documented setup supports CLI, self-hosted and server-oriented use, with Python 3.10 or newer required for the Python installation path.

  • The framework is Python-based and provides both a CLI and Python SDK.
  • MCP integration is documented as a native capability.
  • Specialized local machine-learning, cheminformatics and visualization tools require extra installation groups.
  • The documentation describes support for long-running asynchronous operations and tool composition.
  • Apple Silicon Macs require macOS 14 or newer for the documented vector-search dependency; the documentation reports no equivalent floor for Intel Macs, Windows or Linux.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the ToolUniverse Python SDK

Create a Python 3.12 virtual environment and install the ToolUniverse SDK.

Before you start

  • Install uv first; the README says not to use system pip.
  • Python 3.10 or newer is required.
  • On Apple Silicon Macs, macOS 14 or newer is required for the documented base dependencies.

Steps

  1. If uv is not installed, install it with the command below.

    Documented command
    curl -LsSf https://astral.sh/uv/install.sh | sh

  2. Create and activate a Python 3.12 virtual environment with the command below.

    Documented command
    uv venv --python 3.12 && source .venv/bin/activate

  3. Install the SDK with the command below.

    Documented command
    uv pip install tooluniverse

Expected outcome

The ToolUniverse SDK is installed in the active virtual environment; the base installation covers API and database tools.

Guide limitations

  • Local ML, cheminformatics, and plotting tools require extras; the README lists optional groups and notes that [all] excludes pdf, singlecell, smolagents, client, and build.
  • Apple Silicon Macs earlier than macOS 14 would need to compile faiss-cpu; Intel Macs, Windows, and Linux have no such floor.
Documentation 1 ↗
Source-based practical guide

Add ToolUniverse to an MCP client configuration

Configure an MCP client to launch ToolUniverse using uvx.

Before you start

  • Access to the MCP client configuration file.
  • uvx available in the environment used to launch the server.

Steps

  1. Add an mcpServers entry named tooluniverse to the MCP configuration file.

  2. Set its command to uvx and its args to ["--refresh", "tooluniverse"].

  3. Set the entry's environment variable PYTHONIOENCODING to utf-8.

  4. Optionally remove --refresh to use uv's cache for faster startup; the README gives the command below as the upgrade step for that cached installation.

    Documented command
    uv cache clean tooluniverse

Expected outcome

The MCP configuration contains a ToolUniverse server entry that launches through uvx.

Guide limitations

  • This recipe covers configuration, not validation or an agent workflow.
  • The README says --refresh checks PyPI for the newest release on every launch; removing it uses uv's cache.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
30 September 2026
Original website title
mims-harvard/ToolUniverse · Official homepage title · retained in its source language
How we review sourcesSuggest a correction