setup-tooluniverse

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Setup ToolUniverse

配置 ToolUniverse

Guide the user step-by-step through setting up ToolUniverse with MCP (Model Context Protocol) integration.

引导用户逐步完成集成MCP（Model Context Protocol）的ToolUniverse安装配置。

Agent Behavior

Agent 行为规范

Be friendly, conversational, and interactive. This is a setup wizard, not a reference manual.

Detect the user's language from their first message. If they write in Chinese, Japanese, Spanish, etc., respond in that language throughout the entire setup. All explanations, questions, and celebrations should be in their language. Only keep commands, code blocks, URLs, and env variable names in English (those are technical and must stay as-is).
Go one step at a time. Don't dump all steps at once.
Ask before proceeding to the next step. Confirm the previous step worked.
Use the AskQuestion tool for structured choices when available (client selection, research areas, etc.).
Explain briefly what each step does and why, in plain language.
When something goes wrong, be reassuring and help troubleshoot before moving on.
Celebrate small wins -- when uv installs successfully, when the MCP server appears, when the first tool call works.

保持友好、口语化且互动性强。这是一个安装配置向导，而非参考手册。

根据用户的第一条消息检测其使用语言。如果用户使用中文、日语、西班牙语等语言，整个配置过程都使用该语言回复。所有说明、问题和祝贺语都使用用户的语言，仅保留命令、代码块、URL和环境变量名称为英文（这些属于技术内容，需保持原样）。
分步进行。不要一次性展示所有步骤。
进入下一步前先确认。确保上一步操作成功后再继续。
在可选时使用AskQuestion工具进行结构化选择（如客户端选择、研究领域等）。
用通俗易懂的语言简要解释每个步骤的作用和原因。
遇到问题时保持耐心，先协助排查再继续。
为小的成功庆祝——比如uv安装成功、MCP服务器启动、首次工具调用成功时。

Internal Notes (do not show to user)

内部说明（请勿展示给用户）

⚠️ ToolUniverse has 1200+ tools which will cause context window overflow if all exposed directly. The default

tooluniverse

command already enables compact mode automatically.

Compact mode exposes only 5 core tools (list_tools, grep_tools, get_tool_info, execute_tool, find_tools) while keeping all 1200+ tools accessible via execute_tool.

⚠️ ToolUniverse包含1200+工具，如果直接全部暴露会导致上下文窗口溢出。默认的

tooluniverse

命令已自动启用紧凑模式。

紧凑模式仅暴露5个核心工具（list_tools、grep_tools、get_tool_info、execute_tool、find_tools），同时通过execute_tool可访问全部1200+工具。

Step 0: Welcome & Discovery

步骤0：欢迎与需求确认

Start by welcoming the user and asking two questions to tailor the setup:

Question 1: Which app are you using?

Use AskQuestion if available:

Cursor
Claude Desktop
VS Code / Copilot
Windsurf
Claude Code
Gemini CLI
Codex (OpenAI)
Cline
Trae
Antigravity
OpenCode
Other

Question 2: How will you use ToolUniverse?

MCP server (use scientific tools through chat) -- this is the default for most users
Python coding (write scripts that
```
import tooluniverse
```
) -- also needs pip install

For MCP-only users, only

uv

is needed.

uvx

automatically installs and runs ToolUniverse.

For coding use, also ask about Python version (

python3 --version

, needs >=3.10, <3.14).

首先欢迎用户，然后通过两个问题定制配置流程：

问题1：您正在使用哪个应用？

如果可用，请使用AskQuestion工具：

Cursor
Claude Desktop
VS Code / Copilot
Windsurf
Claude Code
Gemini CLI
Codex (OpenAI)
Cline
Trae
Antigravity
OpenCode
其他

问题2：您将如何使用ToolUniverse？

MCP服务器（通过聊天使用科研工具）——这是大多数用户的默认选择
Python编码（编写
```
import tooluniverse
```
的脚本）——还需要pip安装

仅使用MCP的用户只需安装

uv

。

uvx

会自动安装并运行ToolUniverse。

对于编码使用场景，还需询问Python版本（执行

python3 --version

，要求>=3.10且<3.14）。

Installation Workflow

安装工作流

Step 1: Make sure uv is installed

步骤1：确保已安装uv

First, check if the user already has

uv

(a fast Python package manager). Run this for them:

bash

which uv || echo "uv not installed"

If it's already installed, let them know and move on. If not, explain that

uv

is a fast package manager that makes MCP setup simple, then install it:

bash

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

Confirm it worked before moving to the next step.

首先检查用户是否已安装

uv

（一款快速的Python包管理器）。请用户执行以下命令：

bash

which uv || echo "uv not installed"

如果已安装，告知用户并进入下一步。如果未安装，说明

uv

是一款能简化MCP配置的快速包管理器，然后执行以下命令安装：

bash

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

确认安装成功后再进入下一步。

Step 2: Add ToolUniverse to your MCP config

步骤2：将ToolUniverse添加到MCP配置

Now we'll add ToolUniverse to your app's MCP configuration. Based on the client the user chose in Step 0, help them find and edit the right config file. All clients use the same server config -- only the file location differs.

现在将ToolUniverse添加到应用的MCP配置中。根据用户在步骤0中选择的客户端，协助其找到并编辑正确的配置文件。所有客户端使用相同的服务器配置——仅文件位置不同。

MCP Server Configuration (same for all clients)

MCP服务器配置（所有客户端通用）

Default method: Using uvx (Recommended)

json

{
  "mcpServers": {
    "tooluniverse": {
      "command": "uvx",
      "args": ["tooluniverse"],
      "env": {
        "PYTHONIOENCODING": "utf-8"
      }
    }
  }
}

uvx tooluniverse

automatically installs and runs ToolUniverse with compact mode enabled by default (the

tooluniverse

entry point enables

--compact-mode

automatically). No separate

pip install

step required.

Alternative: Using a pre-installed command (if user already has ToolUniverse installed via pip)

json

{
  "mcpServers": {
    "tooluniverse": {
      "command": "tooluniverse",
      "args": [],
      "env": {
        "PYTHONIOENCODING": "utf-8"
      }
    }
  }
}

默认方法：使用uvx（推荐）

json

{
  "mcpServers": {
    "tooluniverse": {
      "command": "uvx",
      "args": ["tooluniverse"],
      "env": {
        "PYTHONIOENCODING": "utf-8"
      }
    }
  }
}

uvx tooluniverse

会自动安装并运行ToolUniverse，且默认启用紧凑模式（

tooluniverse

入口点会自动启用

--compact-mode

），无需单独执行

pip install

步骤。

替代方法：使用预安装命令（如果用户已通过pip安装ToolUniverse）

json

{
  "mcpServers": {
    "tooluniverse": {
      "command": "tooluniverse",
      "args": [],
      "env": {
        "PYTHONIOENCODING": "utf-8"
      }
    }
  }
}

Config File Locations by Client

各客户端的配置文件位置

Client	Config File (macOS)	How to Access
Cursor	`~/.cursor/mcp.json`	Settings → MCP → Add new global MCP server
Claude Desktop	`~/Library/Application Support/Claude/claude_desktop_config.json`	Settings → Developer → Edit Config
Claude Code	`~/.claude.json` (user) or `.mcp.json` (project)	`claude mcp add` CLI or edit directly
Windsurf	`~/.codeium/windsurf/mcp_config.json`	Click MCP hammer icon → Configure
VS Code (Copilot)	`.vscode/mcp.json` (workspace) or user profile `mcp.json`	Cmd Palette → "MCP: Add Server" (see different format below)
Cline	`cline_mcp_settings.json` (in VS Code extension globalStorage)	Cline panel → MCP Servers → Configure
Codex (OpenAI)	`~/.codex/config.toml`	Create/edit manually (TOML format, see below)
Gemini CLI	`~/.gemini/settings.json` (user) or `.gemini/settings.json` (project)	`gemini mcp add` CLI or edit directly
Antigravity	`mcp_config.json`	"..." dropdown → Manage MCP Servers → View raw config
Trae	`.trae/mcp.json` (project) or global via Trae UI	Ctrl+U → AI Management → MCP → Configure Manually
OpenCode	`~/.config/opencode/opencode.json` or `opencode.json` (project)	Edit directly

Windows/Linux paths differ -- check your client's documentation for the exact location.

Most clients use the same JSON

mcpServers

format shown above. Exceptions: VS Code uses

"servers"

key, Codex uses TOML, and OpenCode uses a

"mcp"

key -- see below for their specific formats.

客户端	配置文件（macOS）	访问方式
Cursor	`~/.cursor/mcp.json`	设置 → MCP → 添加新的全局MCP服务器
Claude Desktop	`~/Library/Application Support/Claude/claude_desktop_config.json`	设置 → 开发者 → 编辑配置
Claude Code	`~/.claude.json` （用户级）或 `.mcp.json` （项目级）	使用 `claude mcp add` 命令行工具或直接编辑
Windsurf	`~/.codeium/windsurf/mcp_config.json`	点击MCP锤子图标 → 配置
VS Code (Copilot)	`.vscode/mcp.json` （工作区）或用户配置中的 `mcp.json`	命令面板 → "MCP: 添加服务器"（格式见下文）
Cline	`cline_mcp_settings.json` （位于VS Code扩展的globalStorage中）	Cline面板 → MCP服务器 → 配置
Codex (OpenAI)	`~/.codex/config.toml`	手动创建/编辑（TOML格式，见下文）
Gemini CLI	`~/.gemini/settings.json` （用户级）或 `.gemini/settings.json` （项目级）	使用 `gemini mcp add` 命令行工具或直接编辑
Antigravity	`mcp_config.json`	"..."下拉菜单 → 管理MCP服务器 → 查看原始配置
Trae	`.trae/mcp.json` （项目级）或通过Trae UI设置全局配置	Ctrl+U → AI管理 → MCP → 手动配置
OpenCode	`~/.config/opencode/opencode.json` 或 `opencode.json` （项目级）	直接编辑

Windows/Linux路径有所不同——请查看对应客户端的文档获取确切位置。

大多数客户端使用上述相同的JSON格式

mcpServers

。例外情况：VS Code使用

"servers"

键，Codex使用TOML格式，OpenCode使用

"mcp"

键——具体格式见下文。

Clients with Different Config Formats

配置格式不同的客户端

VS Code (Copilot) -- uses

"servers"

key (not

"mcpServers"

) and requires

"type"

field. Add to

.vscode/mcp.json

json

{
  "servers": {
    "tooluniverse": {
      "type": "stdio",
      "command": "uvx",
      "args": ["tooluniverse"],
      "env": { "PYTHONIOENCODING": "utf-8" }
    }
  }
}

Codex (TOML format) -- add to

~/.codex/config.toml

toml

[mcp_servers.tooluniverse]
command = "uvx"
args = ["tooluniverse"]
env = { "PYTHONIOENCODING" = "utf-8" }

OpenCode -- uses

mcp

key with

type

and

command

as array in

opencode.json

json

{
  "mcp": {
    "tooluniverse": {
      "type": "local",
      "command": ["uvx", "tooluniverse"],
      "enabled": true,
      "environment": { "PYTHONIOENCODING": "utf-8" }
    }
  }
}

VS Code (Copilot)——使用

"servers"

键（而非

"mcpServers"

），且需要

"type"

字段。添加到

.vscode/mcp.json

：

json

{
  "servers": {
    "tooluniverse": {
      "type": "stdio",
      "command": "uvx",
      "args": ["tooluniverse"],
      "env": { "PYTHONIOENCODING": "utf-8" }
    }
  }
}

Codex (TOML格式)——添加到

~/.codex/config.toml

：

toml

[mcp_servers.tooluniverse]
command = "uvx"
args = ["tooluniverse"]
env = { "PYTHONIOENCODING" = "utf-8" }

OpenCode——在

opencode.json

中使用

mcp

键，

type

和

command

为数组格式：

json

{
  "mcp": {
    "tooluniverse": {
      "type": "local",
      "command": ["uvx", "tooluniverse"],
      "enabled": true,
      "environment": { "PYTHONIOENCODING": "utf-8" }
    }
  }
}

Step 3 (Only if user chose coding use): Install Python package

步骤3（仅适用于编码使用场景）：安装Python包

Skip this if the user only needs MCP. For coding use, install into their Python environment:

bash

pip install tooluniverse

Then verify together:

python

from tooluniverse import ToolUniverse
tu = ToolUniverse()
print(f"ToolUniverse version: {tu.__version__}")

Let them know this is separate from the MCP server --

uvx

installs into uv's cache, while

pip install

puts it in their Python environment for importing.

如果用户仅使用MCP，可跳过此步骤。对于编码使用场景，将ToolUniverse安装到Python环境中：

bash

pip install tooluniverse

然后共同验证安装：

python

from tooluniverse import ToolUniverse
tu = ToolUniverse()
print(f"ToolUniverse version: {tu.__version__}")

告知用户这与MCP服务器是分开的——

uvx

会将ToolUniverse安装到uv的缓存中，而

pip install

会将其安装到Python环境中以便导入使用。

Step 4: Set up API Keys

步骤4：配置API密钥

Many tools work without API keys, but some unlock powerful features. Before diving into keys, ask the user about their research interests to recommend only what's relevant.

Read API_KEYS_REFERENCE.md for detailed per-key info (what it does, step-by-step registration, which tools need it).

许多工具无需API密钥即可使用，但部分工具需要密钥才能解锁强大功能。在配置密钥前，询问用户的研究兴趣，以便仅推荐相关的密钥。

查看API_KEYS_REFERENCE.md获取各密钥的详细信息（功能、分步注册流程、适用工具）。

How to guide API key setup

如何引导用户配置API密钥

Ask the user what research areas they're interested in. Use AskQuestion if available with options like:
- Literature search & publications
- Drug discovery & pharmacology
- Protein structure & interactions
- Genomics & disease associations
- Rare diseases & clinical
- Enzymology & biochemistry
- Patent search
- AI-powered analysis (needs LLM key)
- All of the above
- Not sure yet / skip for now
Map their answer to recommended keys using the tiers below. Don't overwhelm -- suggest 2-4 keys to start.
Walk through each key one at a time:
- Explain in plain language what it unlocks (e.g., "This lets you search PubMed faster")
- Give them the registration link
- Wait for them to sign up and get the key
- Help them add it to their config file
- Move to the next key
After all keys are added, restart the app and test one key with a real tool call.
Let them know they can always come back to add more keys later.

询问用户的研究领域。如果可用，使用AskQuestion工具提供以下选项：
- 文献检索与出版物
- 药物发现与药理学
- 蛋白质结构与相互作用
- 基因组学与疾病关联
- 罕见病与临床研究
- 酶学与生物化学
- 专利检索
- AI驱动的分析（需要LLM密钥）
- 以上全部
- 暂时不确定/跳过
根据用户的回答映射到推荐的密钥，参考以下层级。不要让用户感到负担过重——建议先配置2-4个密钥。
逐个引导用户配置密钥：
- 用通俗易懂的语言解释该密钥的作用（例如："此密钥可让您更快地检索PubMed文献"）
- 提供注册链接
- 等待用户注册并获取密钥
- 协助用户将密钥添加到配置文件中
- 继续配置下一个密钥
所有密钥配置完成后，重启应用并通过实际工具调用测试其中一个密钥。
告知用户以后随时可以回来添加更多密钥。

Tier 1: Core Scientific Keys (Recommended for most users)

层级1：核心科研密钥（推荐大多数用户配置）

Key	Service	What It Unlocks	Free?	Registration
`NCBI_API_KEY`	NCBI/PubMed	PubMed literature search (raises rate limit 3->10 req/s)	Yes	https://account.ncbi.nlm.nih.gov/settings/
`NVIDIA_API_KEY`	NVIDIA NIM	16 tools: AlphaFold2 structure prediction, molecular docking, genomics	Yes	https://build.nvidia.com
`BIOGRID_API_KEY`	BioGRID	Protein-protein interaction queries	Yes	https://webservice.thebiogrid.org/
`DISGENET_API_KEY`	DisGeNET	5 gene-disease association tools	Yes (academic)	https://disgenet.com/academic-apply

密钥	服务	解锁功能	免费？	注册链接
`NCBI_API_KEY`	NCBI/PubMed	PubMed文献检索（将请求速率限制从3次/秒提升至10次/秒）	是	https://account.ncbi.nlm.nih.gov/settings/
`NVIDIA_API_KEY`	NVIDIA NIM	16个工具：AlphaFold2结构预测、分子对接、基因组学工具	是	https://build.nvidia.com
`BIOGRID_API_KEY`	BioGRID	蛋白质-蛋白质相互作用查询	是	https://webservice.thebiogrid.org/
`DISGENET_API_KEY`	DisGeNET	5个基因-疾病关联工具	是（学术用途）	https://disgenet.com/academic-apply

Tier 2: Specialized Scientific Keys (Based on research needs)

层级2：专业科研密钥（根据研究需求选择）

Key	Service	What It Unlocks	Free?	Registration
`OMIM_API_KEY`	OMIM	4 Mendelian/rare disease tools	Yes	https://omim.org/api
`ONCOKB_API_TOKEN`	OncoKB	Precision oncology annotations	Yes (academic)	https://www.oncokb.org/apiAccess
`UMLS_API_KEY`	UMLS/NLM	5 medical terminology & concept mapping tools	Yes	https://uts.nlm.nih.gov/uts/
`USPTO_API_KEY`	USPTO	6 patent search & analysis tools	Yes	https://account.uspto.gov/api-manager/
`SEMANTIC_SCHOLAR_API_KEY`	Semantic Scholar	Literature search (raises rate limit 1->100 req/s)	Yes	https://www.semanticscholar.org/product/api
`FDA_API_KEY`	openFDA	Drug/food/device adverse event queries (raises limit 240->1000 req/min)	Yes	https://open.fda.gov/apis/authentication/
`BRENDA_EMAIL` + `BRENDA_PASSWORD`	BRENDA	3 enzyme database tools (both email and password required)	Yes	https://brenda-enzymes.org/register.php

密钥	服务	解锁功能	免费？	注册链接
`OMIM_API_KEY`	OMIM	4个孟德尔/罕见病工具	是	https://omim.org/api
`ONCOKB_API_TOKEN`	OncoKB	精准肿瘤学注释	是（学术用途）	https://www.oncokb.org/apiAccess
`UMLS_API_KEY`	UMLS/NLM	5个医学术语与概念映射工具	是	https://uts.nlm.nih.gov/uts/
`USPTO_API_KEY`	USPTO	6个专利检索与分析工具	是	https://account.uspto.gov/api-manager/
`SEMANTIC_SCHOLAR_API_KEY`	Semantic Scholar	文献检索（将请求速率限制从1次/秒提升至100次/秒）	是	https://www.semanticscholar.org/product/api
`FDA_API_KEY`	openFDA	药物/食品/器械不良事件查询（将请求速率限制从240次/分钟提升至1000次/分钟）	是	https://open.fda.gov/apis/authentication/
`BRENDA_EMAIL` + `BRENDA_PASSWORD`	BRENDA	3个酶数据库工具（需要同时提供邮箱和密码）	是	https://brenda-enzymes.org/register.php

Tier 3: LLM Provider Keys (For agentic tool features)

层级3：LLM提供商密钥（用于智能工具功能）

At least one LLM key is needed for agentic features. The system tries providers in order: Azure OpenAI -> OpenRouter -> Gemini.

Key	Service	What It Unlocks	Free Tier?	Registration
`GEMINI_API_KEY`	Google Gemini	Agentic tools via Gemini (good free tier)	Yes	https://aistudio.google.com/apikey
`OPENROUTER_API_KEY`	OpenRouter	Agentic tools via 100+ LLM models	Pay-per-use	https://openrouter.ai/
`OPENAI_API_KEY`	OpenAI	Embedding features, LLM-based tool finding	Pay-per-use	https://platform.openai.com/
`AZURE_OPENAI_API_KEY`	Azure OpenAI	Agentic tools via Azure (enterprise)	Pay-per-use	Azure Portal
`ANTHROPIC_API_KEY`	Anthropic Claude	Claude-based features	Pay-per-use	https://console.anthropic.com/
`HF_TOKEN`	HuggingFace	Model/dataset access, HF Inference API	Yes	https://huggingface.co/settings/tokens

至少需要一个LLM密钥才能使用智能工具功能。系统会按以下顺序尝试提供商：Azure OpenAI -> OpenRouter -> Gemini。

密钥	服务	解锁功能	免费层级？	注册链接
`GEMINI_API_KEY`	Google Gemini	通过Gemini使用智能工具（免费层级友好）	是	https://aistudio.google.com/apikey
`OPENROUTER_API_KEY`	OpenRouter	通过100+ LLM模型使用智能工具	按使用量付费	https://openrouter.ai/
`OPENAI_API_KEY`	OpenAI	嵌入功能、基于LLM的工具检索	按使用量付费	https://platform.openai.com/
`AZURE_OPENAI_API_KEY`	Azure OpenAI	通过Azure使用智能工具（企业级）	按使用量付费	Azure Portal
`ANTHROPIC_API_KEY`	Anthropic Claude	基于Claude的功能	按使用量付费	https://console.anthropic.com/
`HF_TOKEN`	HuggingFace	模型/数据集访问、HF推理API	是	https://huggingface.co/settings/tokens

Adding Keys to Configuration

将密钥添加到配置中

Best approach: Add to the
env
block in your MCP config file (the same file from Step 2). This way keys are passed directly to the MCP server:

json

"env": {
  "PYTHONIOENCODING": "utf-8",
  "NCBI_API_KEY": "your_key_here",
  "NVIDIA_API_KEY": "your_key_here"
}

Alternative: Create a

.env

file in your project directory with

KEY=value

pairs.

After adding keys: Restart the app for changes to take effect.

最佳方式：添加到MCP配置文件的
env
块中（即步骤2中的配置文件）。这样密钥会直接传递给MCP服务器：

json

"env": {
  "PYTHONIOENCODING": "utf-8",
  "NCBI_API_KEY": "your_key_here",
  "NVIDIA_API_KEY": "your_key_here"
}

替代方式：在项目目录中创建

.env

文件，写入

KEY=value

格式的内容。

添加密钥后：重启应用使更改生效。

Verify keys work

验证密钥是否生效

Test each configured key with a real tool call:

NCBI_API_KEY

execute_tool("PubMed_search_articles", {"query": "CRISPR", "max_results": 1})

NVIDIA_API_KEY

execute_tool("NvidiaNIM_alphafold2_predict", {"sequence": "MKTVRQERLKS"})

BIOGRID_API_KEY

execute_tool("BioGRID_get_interactions", {"geneList": "TP53", "taxId": 9606})

通过实际工具调用测试每个已配置的密钥：

NCBI_API_KEY

execute_tool("PubMed_search_articles", {"query": "CRISPR", "max_results": 1})

NVIDIA_API_KEY

execute_tool("NvidiaNIM_alphafold2_predict", {"sequence": "MKTVRQERLKS"})

BIOGRID_API_KEY

execute_tool("BioGRID_get_interactions", {"geneList": "TP53", "taxId": 9606})

Step 5: Test it together

步骤5：共同测试

Time to see it in action! Ask the user to restart their app, then try a real tool call together:

Restart the app (Cursor/Claude Desktop/etc.) so the new MCP config is picked up
Check the MCP server appears in the server list

Run a test call -- suggest something relevant to their research interests:

General:
```
grep_tools
```
with keyword "protein" or
```
list_tools
```

Literature:

execute_tool("PubMed_search_articles", {"query": "CRISPR", "max_results": 1})

Drug discovery:

execute_tool("ChEMBL_search_compound", {"query": "aspirin"})

If it works, celebrate and move to Step 6. If something goes wrong, check the Common Issues section below.

现在验证配置是否生效！请用户重启应用，然后共同尝试一次实际的工具调用：

重启应用（Cursor/Claude Desktop等），使新的MCP配置生效
检查MCP服务器是否出现在服务器列表中
执行测试调用——根据用户的研究兴趣推荐合适的调用：
- 通用测试：使用
```
grep_tools
```
  搜索关键词"protein"，或执行
```
list_tools
```
- 文献研究：
```
execute_tool("PubMed_search_articles", {"query": "CRISPR", "max_results": 1})
```
- 药物发现：
```
execute_tool("ChEMBL_search_compound", {"query": "aspirin"})
```

如果调用成功，庆祝并进入步骤6。如果出现问题，请查看下文的常见问题部分。

Step 6: Install ToolUniverse Skills (Highly Recommended)

步骤6：安装ToolUniverse Skills（强烈推荐）

Strongly recommend this step. Skills are pre-built research workflows that teach the AI how to conduct comprehensive scientific research -- they turn basic tool calls into expert-level investigations.

Explain to the user: "ToolUniverse comes with 19 research skills that act like expert guides. For example, the drug-research skill knows exactly which tools to call, in what order, to build a complete drug profile. Without skills, you'd need to figure out which of the 1200+ tools to use yourself."

强烈推荐完成此步骤。Skills是预构建的研究工作流，可指导AI如何开展全面的科研工作——它们能将基础的工具调用转化为专家级的研究。

向用户解释："ToolUniverse包含19个研究Skill，相当于专家向导。例如，药物研究Skill知道应该调用哪些工具以及调用顺序，以生成完整的药物概况。如果没有Skill，您需要自己从1200+工具中选择合适的工具。"

Available Skills

可用的Skills

Skill	What It Does
`tooluniverse`	General strategies for using 1200+ tools effectively
`tooluniverse-drug-research`	Comprehensive drug profiling (identity, pharmacology, safety, ADMET)
`tooluniverse-target-research`	Drug target intelligence (structure, interactions, druggability)
`tooluniverse-disease-research`	Systematic disease analysis across 10 research dimensions
`tooluniverse-literature-deep-research`	Thorough literature reviews with evidence grading
`tooluniverse-drug-repurposing`	Find new therapeutic uses for existing drugs
`tooluniverse-precision-oncology`	Mutation-based treatment recommendations for cancer
`tooluniverse-rare-disease-diagnosis`	Phenotype-to-diagnosis for suspected rare diseases
`tooluniverse-pharmacovigilance`	Drug safety signal analysis from FDA adverse event data
`tooluniverse-infectious-disease`	Rapid pathogen characterization & drug repurposing
`tooluniverse-protein-structure-retrieval`	Protein 3D structure retrieval & quality assessment
`tooluniverse-sequence-retrieval`	DNA/RNA/protein sequence retrieval from NCBI/ENA
`tooluniverse-chemical-compound-retrieval`	Chemical compound data from PubChem/ChEMBL
`tooluniverse-expression-data-retrieval`	Gene expression & omics datasets
`tooluniverse-variant-interpretation`	Genetic variant clinical interpretation
`tooluniverse-protein-therapeutic-design`	AI-guided protein therapeutic design
`tooluniverse-binder-discovery`	Small molecule binder discovery via virtual screening
`tooluniverse-sdk`	Build research pipelines with the Python SDK
`setup-tooluniverse`	This setup guide

Skill	功能
`tooluniverse`	有效使用1200+工具的通用策略
`tooluniverse-drug-research`	全面的药物分析（身份、药理学、安全性、ADMET）
`tooluniverse-target-research`	药物靶点情报（结构、相互作用、成药性）
`tooluniverse-disease-research`	跨10个研究维度的系统性疾病分析
`tooluniverse-literature-deep-research`	带有证据分级的深度文献综述
`tooluniverse-drug-repurposing`	为现有药物寻找新的治疗用途
`tooluniverse-precision-oncology`	基于突变的癌症治疗建议
`tooluniverse-rare-disease-diagnosis`	基于表型的罕见病诊断
`tooluniverse-pharmacovigilance`	基于FDA不良事件数据的药物安全信号分析
`tooluniverse-infectious-disease`	快速病原体特征分析与药物重定位
`tooluniverse-protein-structure-retrieval`	蛋白质3D结构检索与质量评估
`tooluniverse-sequence-retrieval`	从NCBI/ENA检索DNA/RNA/蛋白质序列
`tooluniverse-chemical-compound-retrieval`	从PubChem/ChEMBL检索化学化合物数据
`tooluniverse-expression-data-retrieval`	基因表达与组学数据集
`tooluniverse-variant-interpretation`	遗传变异的临床解读
`tooluniverse-protein-therapeutic-design`	AI引导的蛋白质治疗药物设计
`tooluniverse-binder-discovery`	通过虚拟筛选发现小分子结合剂
`tooluniverse-sdk`	使用Python SDK构建研究管道
`setup-tooluniverse`	本安装配置指南

How to Install Skills

如何安装Skills

Skills are in the

skills/

folder of the ToolUniverse GitHub repo: https://github.com/mims-harvard/ToolUniverse/tree/main/skills

First, help the user download the skills:

bash

undefined

Skills位于ToolUniverse GitHub仓库的

skills/

文件夹中：https://github.com/mims-harvard/ToolUniverse/tree/main/skills

首先，协助用户下载Skills：

bash

undefined

Clone just the skills folder from GitHub

仅从GitHub克隆skills文件夹

git clone --depth 1 --filter=blob:none --sparse https://github.com/mims-harvard/ToolUniverse.git /tmp/tu-skills cd /tmp/tu-skills && git sparse-checkout set skills


Then install based on their client:

**Cursor** -- copy to `.cursor/skills/` (auto-discovered):
```bash
mkdir -p .cursor/skills && cp -r /tmp/tu-skills/skills/* .cursor/skills/

Windsurf -- Cascade supports SKILL.md files the same way:

bash

mkdir -p .windsurf/skills && cp -r /tmp/tu-skills/skills/* .windsurf/skills/

Codex (OpenAI) -- uses

.agents/skills/

directory (auto-discovered):

bash

mkdir -p .agents/skills && cp -r /tmp/tu-skills/skills/* .agents/skills/

Gemini CLI -- uses

.gemini/skills/

directory (auto-discovered):

bash

mkdir -p .gemini/skills && cp -r /tmp/tu-skills/skills/* .gemini/skills/

Claude Code -- uses

.claude/skills/

directory (auto-discovered):

bash

mkdir -p .claude/skills && cp -r /tmp/tu-skills/skills/* .claude/skills/

OpenCode -- uses

.opencode/skills/

directory (auto-discovered):

bash

mkdir -p .opencode/skills && cp -r /tmp/tu-skills/skills/* .opencode/skills/

Trae -- supports skills via

.trae/skills/

bash

mkdir -p .trae/skills && cp -r /tmp/tu-skills/skills/* .trae/skills/

Cline / VS Code / Qwen -- copy skills into the project and reference as needed:

bash

mkdir -p .skills && cp -r /tmp/tu-skills/skills/* .skills/

Clean up after copying:

bash

rm -rf /tmp/tu-skills

git clone --depth 1 --filter=blob:none --sparse https://github.com/mims-harvard/ToolUniverse.git /tmp/tu-skills cd /tmp/tu-skills && git sparse-checkout set skills


然后根据用户的客户端进行安装：

**Cursor**——复制到`.cursor/skills/`（会自动识别）：
```bash
mkdir -p .cursor/skills && cp -r /tmp/tu-skills/skills/* .cursor/skills/

Windsurf——Cascade以相同方式支持SKILL.md文件：

bash

mkdir -p .windsurf/skills && cp -r /tmp/tu-skills/skills/* .windsurf/skills/

Codex (OpenAI)——使用

.agents/skills/

目录（会自动识别）：

bash

mkdir -p .agents/skills && cp -r /tmp/tu-skills/skills/* .agents/skills/

Gemini CLI——使用

.gemini/skills/

目录（会自动识别）：

bash

mkdir -p .gemini/skills && cp -r /tmp/tu-skills/skills/* .gemini/skills/

Claude Code——使用

.claude/skills/

目录（会自动识别）：

bash

mkdir -p .claude/skills && cp -r /tmp/tu-skills/skills/* .claude/skills/

OpenCode——使用

.opencode/skills/

目录（会自动识别）：

bash

mkdir -p .opencode/skills && cp -r /tmp/tu-skills/skills/* .opencode/skills/

Trae——通过

.trae/skills/

支持Skills：

bash

mkdir -p .trae/skills && cp -r /tmp/tu-skills/skills/* .trae/skills/

Cline / VS Code / Qwen——将Skills复制到项目中并按需引用：

bash

mkdir -p .skills && cp -r /tmp/tu-skills/skills/* .skills/

复制完成后清理临时文件：

bash

rm -rf /tmp/tu-skills

How to Use Skills

如何使用Skills

Explain to the user:

Skills activate automatically when the AI detects a relevant request
The user can also trigger them explicitly, e.g., "Research the drug aspirin" will activate the drug-research skill
Skills guide the AI through multi-step research workflows, calling the right tools in the right order
The output is typically a comprehensive research report with evidence grading and source citations

Suggest a skill to try based on their research interests from Step 4. For example:

"Try asking: 'Research the drug metformin' -- the drug-research skill will generate a full drug profile"
"Try asking: 'What does the literature say about CRISPR in cancer?' -- the literature-deep-research skill will do a thorough review"

向用户解释：

当AI检测到相关请求时，Skills会自动激活
用户也可以显式触发Skills，例如："研究药物阿司匹林"会激活药物研究Skill
Skills会引导AI完成多步骤研究工作流，按正确顺序调用合适的工具
输出通常是带有证据分级和来源引用的全面研究报告

根据用户的研究兴趣建议尝试一个Skill，例如：

"试试问：'研究药物二甲双胍'——药物研究Skill会生成完整的药物概况"
"试试问：'文献中关于CRISPR在癌症中的应用有哪些？'——深度文献研究Skill会进行全面综述"

Common Issues & Solutions

常见问题与解决方案

Issue 1: Python Version Incompatibility

问题1：Python版本不兼容

Symptom:

requires-python = ">=3.10"

error. Fix:

brew install python@3.12

then

python3.12 -m pip install tooluniverse

症状：出现

requires-python = ">=3.10"

错误。解决方法：执行

brew install python@3.12

，然后执行

python3.12 -m pip install tooluniverse

Issue 2: uvx or uv Not Found

问题2：找不到uvx或uv

Symptom:

uvx: command not found

. Fix:

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

then restart shell (

source ~/.zshrc

)

症状：出现

uvx: command not found

错误。解决方法：执行

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

，然后重启shell（

source ~/.zshrc

）

Issue 3: Context Window Overflow

问题3：上下文窗口溢出

Symptom: MCP server loads but Cursor/Claude becomes slow or errors

Solution: Enable compact mode (should already be set):

Verify
```
--compact-mode
```
is in args
Restart application
Check you're using stdio command, not HTTP server

症状：MCP服务器已加载，但Cursor/Claude变得缓慢或出现错误

解决方案：启用紧凑模式（默认已开启）：

验证参数中包含
```
--compact-mode
```
重启应用
确认使用的是stdio命令，而非HTTP服务器

Issue 4: Import Errors for Specific Tools

问题4：特定工具导入错误

Symptom: Some tools fail with

ModuleNotFoundError

Solution: Install optional dependencies:

pip install tooluniverse[all]

(or specific extras like

[singlecell]

[ml,embedding]

[visualization]

)

症状：部分工具调用失败，出现

ModuleNotFoundError

错误

解决方案：安装可选依赖：

pip install tooluniverse[all]

（或特定的扩展依赖，如

[singlecell]

、

[ml,embedding]

、

[visualization]

）

Issue 5: MCP Server Won't Start

问题5：MCP服务器无法启动

Symptom: No tooluniverse server appears in Cursor/Claude

Diagnostic steps:

Test command directly:

bash

uvx tooluniverse
# Should start without errors (Ctrl+C to exit)

Check JSON syntax in mcp.json (no trailing commas, proper quotes)

View Cursor logs:

~/Library/Application Support/Cursor/logs/

(macOS)

Verify file permissions on mcp.json

症状：Cursor/Claude中未显示tooluniverse服务器

诊断步骤：

直接测试命令：

bash

uvx tooluniverse
# 应无错误启动（按Ctrl+C退出）

检查mcp.json的JSON语法（无 trailing commas、引号格式正确）

查看Cursor日志：

~/Library/Application Support/Cursor/logs/

（macOS）

验证mcp.json的文件权限

Issue 6: API Key Errors (401/403)

问题6：API密钥错误（401/403）

Symptom: Tool returns "unauthorized", "forbidden", or "invalid API key"

Common causes and fixes:

Key not set: Check it's in the
```
env
```
block of mcp.json (or
```
.env
```
file) and app was restarted
Wrong key name: Double-check the exact env variable name (e.g.,
```
ONCOKB_API_TOKEN
```
not
```
ONCOKB_API_KEY
```
)
Key expired or revoked: Log into the service and regenerate the key
Copy-paste error: Make sure there are no extra spaces or newlines in the key value
Free tier limits: Some services (DisGeNET, OMIM) require account approval before the key works

症状：工具返回"unauthorized"、"forbidden"或"invalid API key"

常见原因与解决方法：

密钥未设置：检查密钥是否已添加到mcp.json的
```
env
```
块（或
```
.env
```
文件）中，且应用已重启
密钥名称错误：仔细核对环境变量的准确名称（例如：
```
ONCOKB_API_TOKEN
```
而非
```
ONCOKB_API_KEY
```
）
密钥过期或被撤销：登录对应服务并重新生成密钥
复制粘贴错误：确保密钥值中没有多余的空格或换行符
免费层级限制：部分服务（如DisGeNET、OMIM）需要账户审核通过后密钥才能生效

Issue 7: Upgrading ToolUniverse

问题7：升级ToolUniverse

Symptom: User wants a newer version, or tools are missing/outdated

uvx

caches packages and may serve an older version. To upgrade:

bash

undefined

症状：用户需要更新版本，或工具缺失/过时

uvx

会缓存包，可能会提供旧版本。如需升级：

bash

undefined

Clear the cached version and get the latest

清理缓存版本并获取最新版本

uv cache clean tooluniverse


Then restart the MCP client. `uvx` will download the latest version on next start.

To pin a specific version:
```json
"args": ["tooluniverse==1.0.18"]

For pip users:

bash

pip install --upgrade tooluniverse

uv cache clean tooluniverse


然后重启MCP客户端。`uvx`会在下次启动时下载最新版本。

如需固定特定版本：
```json
"args": ["tooluniverse==1.0.18"]

对于使用pip的用户：

bash

pip install --upgrade tooluniverse

Still stuck?

仍无法解决？

If you run into an issue that can't be resolved with the steps above, encourage the user to open a GitHub issue at https://github.com/mims-harvard/ToolUniverse/issues or reach out to Shanghua Gao, the creator of ToolUniverse.

如果遇到上述步骤无法解决的问题，鼓励用户在GitHub提交issue：https://github.com/mims-harvard/ToolUniverse/issues，或联系ToolUniverse的创建者[Shanghua Gao](mailto:shanghuagao@gmail.com)。

What's Next?

下一步建议

After setup is complete, suggest the user try one of these to get started:

"Research the drug metformin" -- triggers the drug-research skill for a full drug profile
"What are the known targets of imatinib?" -- triggers target-research
"What does the literature say about CRISPR in sickle cell disease?" -- triggers literature-deep-research
"Find protein structures for human EGFR" -- triggers protein-structure-retrieval

Point them to the tooluniverse
general skill for tips on getting the most out of 1200+ tools, and remind them they can always come back to add more API keys or skills later.

配置完成后，建议用户尝试以下操作快速上手：

"研究药物二甲双胍"——触发药物研究Skill，生成完整的药物概况
"伊马替尼的已知靶点有哪些？"——触发靶点研究Skill
"文献中关于CRISPR在镰状细胞病中的应用有哪些？"——触发深度文献研究Skill
"查找人类EGFR的蛋白质结构"——触发蛋白质结构检索Skill

引导用户查看**

tooluniverse

通用Skill**，获取有效使用1200+工具的技巧，并提醒用户以后随时可以回来添加更多API密钥或Skills。

Quick Reference

快速参考

Default setup:
```
uvx tooluniverse
```
-- auto-installs, auto-enables compact mode
Upgrade:
```
uv cache clean tooluniverse
```
then restart the app
First load: May take 30-60 seconds (downloads + installs); subsequent loads are fast
All scientific API keys are free to obtain
Agentic features need at least one LLM key (Gemini has a good free tier)
Detailed API key docs: API_KEYS_REFERENCE.md
Skills repo: https://github.com/mims-harvard/ToolUniverse/tree/main/skills
Need help? Open a GitHub issue or email Shanghua Gao

默认配置：
```
uvx tooluniverse
```
——自动安装、自动启用紧凑模式
升级方法：
```
uv cache clean tooluniverse
```
，然后重启应用
首次加载：可能需要30-60秒（下载+安装）；后续加载速度很快
所有科研API密钥均可免费获取
智能工具功能需要至少一个LLM密钥（Gemini的免费层级友好）
详细API密钥文档：API_KEYS_REFERENCE.md
Skills仓库：https://github.com/mims-harvard/ToolUniverse/tree/main/skills
需要帮助？ 提交GitHub issue或发送邮件至Shanghua Gao