What this resource does
Core uses
This toolkit combines scientific tools, research Skills, and several access methods
ToolUniverse exposes scientific tools through an MCP server, a command-line interface, a Python SDK, and Agent Skills. This page shows representative research Skills and setup components rather than the complete tool registry.
The official documentation describes 68 pre-built research workflows and 1000+ scientific tools; the cards below are representative.
Scientific literature and knowledge tools
Provides unified access to literature search and scientific knowledge sources including PubMed, Semantic Scholar, arXiv, bioRxiv, and Europe PMC.
- Typical inputs
- Research query, identifier, or entity; Selected search tools and source scope
- Output
- Literature records, source metadata, or structured scientific knowledge results.
Best for: Finding scientific evidence across multiple documented sources.
Components for this task
tooluniverse-literature-deep-research
Runs literature review workflows across PubMed, Europe PMC, bioRxiv, and citation networks with evidence grading.
Suggested requests (1)
What does the literature say about CRISPR in cancer?
From plugin/skills/setup-tooluniverse/SKILL.md, demo query table.
tooluniverse-dataset-discovery
Finds and evaluates research datasets by mapping a question to suitable study designs and repositories.
Suggested requests (2)
Find research data about [topic].
Where can I get a dataset for [research question or topic]?
Adapted into complete requests from official trigger wording in plugin/skills/tooluniverse-dataset-discovery/SKILL.md frontmatter.
Biomedical and scientific analysis
Includes Skills and tools for omics, genetics, drug discovery, protein structure, epidemiology, statistics, data wrangling, and modelling.
- Typical inputs
- Domain dataset, identifiers, and analysis question; Selected Skill or ToolUniverse tool
- Output
- Component-dependent analysis results, tables, model outputs, or structured reports.
Best for: Running a documented scientific tool or domain workflow through one interface.
Components for this task
tooluniverse-data-integration-analysis
Adds gene, pathway, disease, and target context to statistical results such as DEGs, GWAS hits, or associations.
tooluniverse-statistical-modeling
Supports documented regression, group-comparison, survival, effect-size, and clinical adverse-event analyses.
Tool discovery and composition
Lets agents discover, inspect, call, cache, and compose scientific tools through MCP, CLI, Python, or Agent Skills.
- Typical inputs
- Research task and tool-selection constraints; Host agent or SDK configuration
- Output
- Tool calls, cached results, and composed workflow artifacts.
Best for: Building or extending an AI scientist system with traceable scientific tools.
Components for this task
setup-tooluniverse
Guides setup for ToolUniverse chat, CLI, and Python SDK access across documented AI clients.
Suggested requests (11)
Research the drug [name].
Research [disease].
What are the known targets of [drug]?
Find protein structures for [protein].
Is [variant] pathogenic?
What drugs could be repurposed for [disease]?
What are the adverse events for [drug]?
Find clinical trials for [disease/drug].
What are the protein interactions for [gene]?
What are the clinical guidelines for [condition]?
Check drug interactions between [drug1] and [drug2].
From plugin/skills/setup-tooluniverse/SKILL.md, Prompt cheat sheet.
tooluniverse-sdk
Provides Python SDK patterns for tool loading, batch execution, search, custom composition, and benchmarking pipelines.
tooluniverse-claude-code-plugin
Installs the Claude Code plugin with the ToolUniverse MCP server, research Skills, commands, hooks, and research agent.
Use boundaries
Limits and checks
Tool and source variability
Availability, evidence quality, licences, and outputs vary by tool.
Inspect the selected tool, source, parameters, and provenance before use.
External data flow and cost
Research inputs may leave the local environment and fees or limits may apply.
Check credentials, provider terms, budgets, and data policy for each tool.
Scientific interpretation
Automated workflows may produce unsupported conclusions.
Validate methods, assumptions, outputs, and source records with domain expertise.
More boundaries
- ToolUniverse does not make every available tool suitable for every scientific domain or study design.
- A tool cannot access restricted datasets, unavailable services, or credentials the user has not legitimately configured.
- Tool results, biomedical interpretations, statistical conclusions, and literature claims require independent expert validation.
Technical details
- Resource type
- Toolkit
- Author or maintainer
- Shanghua Gao and the ToolUniverse contributors
- Version
- v1.3.1
- Latest release
- v1.3.1
- Source last updated
- 11 Jul 2026
- Last verified
- 13 Jul 2026
- Documented applications
- General Agent Skills, OpenAI API, Anthropic API, MCP-compatible agents, ToolUniverse Python SDK
- Documented AI models
- Claude, GPT, Gemini, Qwen, DeepSeek, Open models
- Licence
- Apache-2.0
- Access
- Publicly available
- Additional costs
- Platform terms or usage limits may apply. API usage fees may apply for selected components. External services, software, compute, or data access may have separate costs.
- Skill instruction language
- English
- Documentation language
- English
- Repository languages
- Python, Markdown
- Dependencies
- uv or Python for the MCP server and SDK; Node.js and npx for Agent Skills installation; Tool-specific APIs, packages, datasets, and credentials
- Review status
- Not tested
Continue exploring