Taixin MCP Server Guide

Why use Taixin MCP

Taixin MCP connects supported AI tools to public product, solution, download, support, and technical-document sources on the Taixin website. It helps engineers retrieve traceable official context without leaving their coding environment.

AI-generated answers can still be inaccurate. Confirm critical specifications, certifications, prices, lead times, and project commitments with the corresponding official document or the Taixin team.

What the Taixin MCP server is

MCP stands for Model Context Protocol. The Taixin server exposes public website search capabilities as MCP tools. It does not provide access to admin APIs, unpublished files, internal systems, credentials, or private customer information.

Server URL

Use the following remote Streamable HTTP endpoint:

https://taixin-semi.com/api/mcp

The first connection opens the Taixin OAuth sign-in and authorization flow. Use any method enabled by the site administrator: Website Email, GitHub, or WeChat. Complete authorization before calling the tools.

Available tools

• search_taixin_sources: search public Taixin website sources for a concrete question.
• get_taixin_product_specs: retrieve public specification sources for a Taixin model.
• find_taixin_downloads: find public datasheets, manuals, SDKs, downloads, and technical documents.
• compare_taixin_products: retrieve public sources for two concrete models to support comparison.

MCP server vs. the website AI assistant

The website AI assistant is used directly in the browser. The MCP server is added to an external client such as Cursor, VS Code, Claude Code, or Codex so the client can retrieve Taixin sources during a working session. Both use public website content and the same security boundaries.

Add to Cursor

Click Add to Cursor in the website AI assistant, allow the browser to open Cursor, and select Install. After installation, select Connect and complete the Taixin MCP sign-in in your browser.

{
  "mcpServers": {
    "taixin-documentation": {
      "url": "https://taixin-semi.com/api/mcp"
    }
  }
}

Add to VS Code

Click Add to VS Code in the website AI assistant, allow the browser to open VS Code, and select Install. Then complete the Taixin MCP sign-in in your browser.

{
  "servers": {
    "taixin-documentation": {
      "type": "http",
      "url": "https://taixin-semi.com/api/mcp"
    }
  }
}

If MCP: List Servers already contains taixin-documentation but the details page still shows Install, this is a UI state issue in some VS Code versions. Do not keep selecting Install. View the server output or open the tool picker to confirm that the server starts and exposes Taixin tools.

You can also use the official VS Code CLI from Windows PowerShell or a macOS/Linux shell:

code --add-mcp '{"name":"taixin-documentation","type":"http","url":"https://taixin-semi.com/api/mcp"}'

Add to Claude Code

Click Add to Claude Code to copy this command, then paste and run it in a terminal:

claude mcp add --transport http --scope user taixin-documentation "https://taixin-semi.com/api/mcp"

After the command succeeds, run claude, enter /mcp, select taixin-documentation, choose Authenticate, and complete sign-in in your browser.

The --scope user option stores the server in user-level ~/.claude.json, so both the Code tab in the Claude Code desktop app and Claude Code CLI can use it. It does not configure the regular chat surface in Claude Desktop.

Configure ChatGPT (Codex) manually

First copy the MCP link from the website AI assistant. If you use Codex CLI, you can also run both commands below. The first registers Taixin MCP and the second starts browser-based OAuth sign-in:

codex mcp add taixin-documentation --url "https://taixin-semi.com/api/mcp"
codex mcp login taixin-documentation

After sign-in, Codex in the ChatGPT desktop app, Codex CLI, and the Codex IDE extension share the user-level MCP configuration in ~/.codex/config.toml. Regular chats on ChatGPT web do not read this local configuration.

Verify the connection

1. Add the server URL and complete the OAuth authorization flow.
2. Confirm that tools/list returns the four Taixin tools.
3. Call get_taixin_product_specs with model TXW8301 and confirm that the response cites Taixin sources.
4. Ask an unrelated generic question and confirm that the tool returns an out-of-scope result without unrelated sources.

Limits

The service only searches published public sources. Content that has not been published or indexed cannot be returned. Requests are also subject to rate limits and abuse protection.

Security and permission boundary

The MCP server does not expose passwords, API keys, admin data, unpublished documents, private customer data, prices, inventory, delivery promises, or internal engineering conclusions. OAuth authorization does not expand the public-content scope.

Troubleshooting

The OAuth page does not open

Check that the client supports remote HTTP MCP servers and OAuth, then remove and add the server again. Corporate proxies or firewalls must allow access to https://taixin-semi.com/api/mcp.

The AI tool does not call Taixin MCP

Confirm that the server is enabled for the current session and that tool use is allowed. You can explicitly ask the client to search official Taixin sources through MCP.

No sources are returned

Include a concrete Taixin model, document type, parameter, or application context. The requested source may not yet be published or indexed.