# Univer SDK MCP

> Connect your coding agent to Univer documentation, API references and initialization tools.

- Human documentation: [https://docs.univer.ai/overview/mcp](https://docs.univer.ai/overview/mcp)

- Agent Markdown: [https://docs.univer.ai/overview/mcp.md](https://docs.univer.ai/overview/mcp.md)

- Requested language: `en-US`

- Content language: `en-US`

- Documentation version: `1.5.0`

- Source: [mcp.mdx](https://github.com/dream-num/documentation/blob/dev/content/docs/mcp.mdx)

---

Univer SDK MCP lets your coding agent search official documentation, read API references and generate browser editor initialization code from the same catalog as the [initialization generator](/tools/initialization-generator).

## Connect your agent

Add this remote server in an agent that supports Streamable HTTP MCP:

| Setting | Value |
| --- | --- |
| Name | `univer-docs` |
| Transport | Streamable HTTP |
| Server URL | `https://docs.univer.ai/mcp/docs` |
| Authentication | No API key required |

After enabling the server, confirm that your agent can see its tools:

| Tool | Purpose |
| --- | --- |
| `search_docs` | Search guides, examples and API references by topic or API name; filter by section and language |
| `get_doc` | Read a page as Markdown with code, source and version information; use `nextLine` to continue long pages |
| `list_features` | Find supported browser feature IDs, availability and license requirements for a product |
| `generate_setup` | Generate dependency and initialization files with plugins, presets, locales and styles |

## Describe your task

> Use Univer SDK MCP to add a spreadsheet to this React application. Inspect the installed SDK version and existing setup first. Read the integration guide, add cell editing and snapshot saving, then build and verify the application.

> Use Univer SDK MCP to find the Facade API for changing a cell's number format. Check the required feature plugin and imports before changing my existing Node.js integration.

Tell your agent the product, framework, runtime and desired behavior. Search snippets are only entry points: the agent should read the relevant guide and API page before implementing. For icons, connect [Univer Icons MCP](https://docs.univer.ai/guides/icons/ai.md).

The service reads the documentation published with this site. Responses identify the documentation or generator version; compare it with the project's installed packages. Generated setups target browser editors, and some features require a license or your backend configuration. For framework lifecycle, Headless Univer, collaboration and file conversion, read the corresponding guides. The server returns file contents; your agent applies and tests them in your project.