Overview
- Skill Key
- hollaugo/mcp-app-builder
- Author
- hollaugo
- Source Repo
- openclaw/skills
- Version
- -
- Source Path
- skills/hollaugo/mcp-app-builder
- Latest Commit SHA
- bbed954b78711ffd43f8a3e463286339eab0f0a1
Build new MCP Apps (MCP servers with React UI output) using @modelcontextprotocol/ext-apps and the MCP SDK. Use when asked to scaffold or implement MCP App servers, add UI-rendering tools/resources, or migrate a standard MCP server to an MCP App with Vite single-file UI bundles.
Stars
0
Installs
0
Status
ACTIVE
Visibility
PUBLIC
直接复制以下提示词,发送给你的 AI 助手即可完成安装。
请先检查是否已安装 SkillHub 商店,若未安装,请根据 https://skillhub-1388575217.cos.ap-guangzhou.myqcloud.com/install/skillhub.md 安装 SkillHub 商店,然后安装 mcp-app-builder 技能。 若已安装,则直接安装 mcp-app-builder 技能。
# MCP App Builder
## Overview
Create MCP Apps that expose tools with visual React UIs for ChatGPT or Claude. Follow the exact dependency versions and server/UI patterns in `references/mcp-app-spec.md`.
## Workflow
1. Clarify requirements: what data to visualize, UI pattern (card, table, chart, dashboard, form), data source, and how many tools (start with 1-2).
2. Design tools and UI mapping: define tool names, zod input schemas, output shape, and UI resource URIs (`ui://.../app.html`). Map each tool to one React entrypoint and one HTML file.
3. Scaffold the project: start from `assets/mcp-app-template/` when possible, then customize tool names, schemas, and UI. Ensure `package.json` uses the exact versions, plus `tsconfig.json`, `vite.config.ts`, Tailwind + PostCSS, and per-tool build scripts.
4. Implement the server: use `registerAppTool`/`registerAppResource`, zod schemas directly, `createServer()` factory per request, and `createMcpExpressApp` with `app.all("/mcp", ...)`.
5. Implement the UI: use `useApp` + `useHostStyles`, parse tool results, handle loading/error/empty states, and apply safe-area insets.
6. Build and test: run `npm run build`, then `npm run serve`, then verify via a tunnel if needed.
## Hard Requirements
- Use the exact dependency versions listed in `references/mcp-app-spec.md`.
- Use `registerAppTool`/`registerAppResource` and zod schemas directly (not JSON Schema objects).
- Create a new `McpServer` instance per request via `createServer()`.
- Use `createMcpExpressApp` and `app.all("/mcp", ...)`.
- Bundle UI into single-file HTML via `vite-plugin-singlefile`.
- Use host CSS variables for theme compatibility.
## References
- `references/mcp-app-spec.md` (authoritative spec, patterns, code templates, gotchas)
## Assets
- `assets/mcp-app-template/` (ready-to-copy MCP App skeleton with one tool + UI)
openstockdata
OpenClaw Skill for stock data analysis
capt-marbles
Generative Engine Optimization (GEO) for AI search visibility. Optimize content to appear in ChatGPT, Perplexity, Claude, and Google AI Overviews. Use when optimizing websites, pages, or content for LLM discoverability and citation.
capgoblin
Access unsecured credit lines for AI agents on the Arc Network using the Credex Protocol. Use for borrowing USDC against reputation, repaying debt to grow credit limits, providing liquidity as an LP, or managing cross-chain USDC via Circle Bridge. Triggers on "borrow from credex", "repay debt", "deposit to pool", "check credit status", "provide liquidity", or any credit/lending task on Arc.
capt-marbles
Control PhantomBuster automation agents via API. List agents, launch automations, get output/results, check status, and abort running agents. Use when the user needs to run LinkedIn scraping, Twitter automation, lead generation phantoms, or any PhantomBuster workflow.
camelsprout
DuckDB CLI specialist for SQL analysis, data processing and file conversion. Use for SQL queries, CSV/Parquet/JSON analysis, database queries, or data conversion. Triggers on "duckdb", "sql", "query", "data analysis", "parquet", "convert data".
camohiddendj
DuckDuckGo HTML search scraper CLI with JSON, CSV, OpenSearch, markdown, and compact outputs.