> For the complete documentation index, see [llms.txt](https://docs.cherryai.com.cn/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.cherryai.com.cn/docs/en-us/advanced-basic/mcp.md). # MCP Usage Tutorial ## Understand MCP in one sentence **MCP is a "unified interface" that lets AI call resources on your computer or online accounts.** Analogy: * By default, AI is like a**phone with no apps installed** — it can only chat, cannot access files on your computer, cannot log into your Notion, and cannot operate a browser * **MCP** like a phone's**app store + universal interface**: by installing an MCP, AI gains a new ability, such as "read local files", "query Notion content", or "send emails on your behalf" MCP was initiated by Anthropic and established as a unified standard. Developers around the world create various tool components according to this standard (called **MCP Server**). Any AI client that supports MCP (including Cherry Studio) can use the same set of tools. ## What can you do with MCP? Here are common scenarios suitable for everyday users: * **Read local files**: let AI access `~/Documents` notes, bills, contracts in * **Connect note-taking tools**: let AI query and modify content in Notion / Obsidian / Apple Notes * **Automatically send and receive messages**: let AI draft and directly send emails and Feishu messages * **Query databases**: let AI automatically write SQL to query MySQL / PostgreSQL * **Browse web pages + screenshots**: let AI operate the browser to obtain information * **Code collaboration**: let AI directly read/modify local code repositories and query GitHub Issues ## The difference between MCP and "skills" The two are easy to confuse; the comparison is as follows: | | **Skill** | **MCP** | | ---------------------------------------------- | ---------------------------------------------------------------------- | ------------------------------------------------------- | | **Core idea** | Ability package for "how to do it" | Interface for "what to use" | | **Analogy** | AI's educational background (way of thinking) | AI's toolbox (operational capability) | | **Typical example** | "Can create Xiaohongshu-style image/text posts", "Can draw flowcharts" | "Can read files", "Can query Notion", "Can send emails" | | **Whether external accounts/tools are needed** | No (model layer only) | Usually yes (integrates external apps / APIs) | **Practical difference**: after installing a "Xiaohongshu skill", AI output will conform to Xiaohongshu style; after installing a Notion MCP, AI can truly access your Notion data. > Recommended reading first [Concepts introduction](/docs/en-us/advanced-basic/concepts-101.md). ## How do I use it? Overall, three steps: 1. **First-time use**:[Install the MCP runtime environment](/docs/en-us/advanced-basic/mcp/install.md)(one-time, 1 minute) 2. **Add an MCP**:[Manual configuration](/docs/en-us/advanced-basic/mcp/config.md) or [Let AI install it automatically](/docs/en-us/advanced-basic/mcp/auto-install.md) 3. **Enable it in the chat/Agent** → Then chat/run tasks as usual, and AI will decide when to invoke it by itself See the three links above for detailed steps. ## Where can I find MCPs? * **the official repository**:[github.com/modelcontextprotocol/servers](https://github.com/modelcontextprotocol/servers)(most complete) * **Cherry Studio built-in marketplace**: In `Settings → MCP Server → Discover → Built-in Servers` Browse and install community and official plugins with one click. * **Cherry Studio built-in service providers**: `Settings → MCP Server → Providers` Some providers are integrated here, and you can use their MCP after configuration. * **Shared by third parties**: Reddit / Discord / personal blogs all have many. ## Common questions **Q: Do I need programming skills to install MCP?** No. Usually you just need to copy a piece of configuration into Cherry Studio, or use [automatic installation](/docs/en-us/advanced-basic/mcp/auto-install.md) and let AI do it for you. **Q: Will a wrong installation damage my computer?** The risk is low. MCP can only perform the operations it declares (one that declares it can read files can only read files; one that does not declare email sending cannot send emails). But**MCPs from unknown sources still pose risks**, so it is recommended to prioritize official repositories. **Q: Will MCP obtain my data?** It depends on the specific MCP implementation. Open-source MCP code in official repositories can be audited and is relatively trustworthy; for third-party closed-source MCPs, please confirm the developer's identity before installing. ## Next step * First-time use: start by reading [MCP environment installation](/docs/en-us/advanced-basic/mcp/install.md) * Want to try one directly: see [Configure and Use MCP](/docs/en-us/advanced-basic/mcp/config.md) follow along once * If manual configuration is too troublesome: try [Automatically Install MCPs](/docs/en-us/advanced-basic/mcp/auto-install.md) * Want to see which MCPs are built into Cherry Studio:[Built-in MCP configuration](/docs/en-us/advanced-basic/mcp/builtin.md) * If you run into problems halfway through installation:[Frequently asked questions](/docs/en-us/advanced-basic/mcp/faq.md) *** ### 💡 Get help and submit feedback If you encounter any questions, bugs, or have suggestions for feature improvements during configuration or use, please refer to [Feedback and Suggestions](/docs/en-us/question-contact/suggestions.md) the official channels provided there. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.cherryai.com.cn/docs/en-us/advanced-basic/mcp.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.