What clients support this server?
Clients must support OAuth2 (with dynamic client registration), tool notifications, and resource notifications. The Portal One web client and MCP Inspector are mentioned as compatible options.
@portal-labs-infrastructure
基本情報
A demonstration Model Context Protocol (MCP) server implemented in TypeScript, showcasing dynamic tool and resource management, stateful HTTP sessions, and clean architectural patterns.
ツール一覧はまだありません
この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。
ツールは README から自動的に抽出されます。メンテナーは ## Tools という見出しの下に記載することで、このタブに反映できます。
A demonstration Model Context Protocol (MCP) server built in TypeScript that hosts a simple number guessing game. It showcases a dynamic, session-based architecture where each connecting user receives their own isolated set of tools and resources managed by a central Express.js controller. Designed as a learning resource and practical example for building dynamic, multi-user MCP applications.
Run the server locally via npm run dev (development) or npm start (production). Clients must support OAuth2 with dynamic client registration, tool notifications, and resource notifications. The live server is accessible at https://mcp.number-guessing-game.portal.one/mcp. For local access, use http://localhost:8083/mcp or expose with ngrok.
start_game, guess_number, give_up) enabled/disabled per user based on game state.game_state) created and destroyed per session.Clients must support OAuth2 (with dynamic client registration), tool notifications, and resource notifications. The Portal One web client and MCP Inspector are mentioned as compatible options.
Node.js v18.x or later, with npm or yarn. Google Cloud Firestore access is needed for state persistence.
A central Express.js controller creates a unique set of MCP tools and resources for each user upon session initialization and destroys them on close, using transport lifecycle hooks (`onsessioninitialized`, `onclose`).
Game state and high scores are persisted in Google Cloud Firestore, making the server stateless and horizontally scalable.
It uses `StreamableHTTPServerTransport` to handle incoming requests at the `/mcp` endpoint.

Search and manage photo and video libraries through AI assistants. Find images and video moments, review duplicates, and organize collections, groups, favorites and people across local folders, Apple Photos and OneDrive.

Break down any TikTok or Instagram Reels ad and rewrite it for your product, right inside Claude, Cursor, or ChatGPT. Hooks, UGC scripts, and CPM. Free, no API key.
Drive your SVGator account from any AI assistant. Prompt an animation, get an editable project back, and export it as SVG, Lottie, GIF, MP4, and more, then open it in the editor to finish by hand.

Generate on brand product videos from Claude Code (or any assistant) in one line. Lamina turns a brief + your brand kit into finished product photos, video reels, and campaign banners - no prompt engineering; brand guard

One workspace of tools for Claude and ChatGPT. Connect apps, generate video, music, and ads, and build new tools on demand. Generate and publish ad creative directly from your assistant.
コメント