// the find
TencentCloudBase/CloudBase-MCP
CloudBase MCP - Connect CloudBase to your AI Agent. Go from AI prompt to live app.
CloudBase MCP is a Model Context Protocol server that bridges AI coding tools (Cursor, Claude Code, etc.) to Tencent's CloudBase platform — their Firebase-like BaaS offering for serverless functions, NoSQL database, static hosting, and WeChat miniprogram deployment. The pitch is that you describe what you want in natural language and the MCP server handles provisioning and deploying the cloud infrastructure without leaving your IDE. It's primarily useful if you're already in the Tencent Cloud ecosystem or building WeChat miniprograms.
The three deployment modes (local npx, Tencent-hosted, self-hosted with cloud mode) are well thought out, and the security isolation in cloud mode — automatically disabling local file system tools — is the right call rather than leaving it to users to figure out. The MCP tool surface is broad: 40+ tools covering functions, database, storage, hosting, miniprogram publishing, and log access, which means a realistic deploy-debug loop can stay inside the AI agent. The skills/rules library distributed with the repo is genuinely useful — AI-native context about CloudBase conventions means generated code is more likely to be correct on the first try rather than requiring manual fixups for platform idioms.
This is tightly coupled to Tencent CloudBase, so if you're not deploying there, there's nothing here for you — unlike generic deployment tools that work across providers. The README is almost entirely in Chinese despite claiming English support, and the English README appears to be a partial translation; the discoverability of specific tool capabilities requires reading source files. Vendor lock-in is real: data models, auth, file storage, and function runtime are all CloudBase-specific, with no migration path documented if you later want to move off. The 'vibe coding' framing in the topics and README language ('from AI prompt to live app') downplays the fact that you still need a funded Tencent Cloud account and need to understand the platform's billing model before anything actually runs in production.