SocialPulse AI

Outils d'analyse et d'automatisation des médias sociaux propulsés par l'IA, conçus avec TensorFlow.js. Ce serveur fournit une suite d'outils d'apprentissage automatique—incluant l'analyse des sentiments, la détection de toxicité, la prédiction d'engagement et le regroupement de contenu—pour analyser les interactions sur plusieurs plateformes comme YouTube, Instagram et Twitter. Il fonctionne localement pour la confidentialité des données et un traitement à faible latence. Fonctionnalités : prise en charge multiplateforme, exécution locale de modèles ML, prêt pour la monétisation x402.

Documentation

TensorFlow.js Social Media MCP Server

MCP Server

A comprehensive suite of machine learning tools for analyzing and automating interactions across multiple social media platforms, built on TensorFlow.js and powered by the Model Context Protocol (MCP).

This server performs ML-based analytics locally (e.g., sentiment analysis, toxicity detection, engagement prediction, content clustering) without relying on external cloud-based AI services, ensuring data privacy and low-latency processing.

Key Features

  • Cross-Platform Support: Tools for YouTube, Instagram, TikTok, Twitter, Facebook, Discord, Twitch, Reddit, LinkedIn, Threads, Bluesky, Mastodon, GitHub, Spotify, and Pinterest.
  • Local Machine Learning: Powered by TensorFlow.js (MobileNet, COCO-SSD, Universal Sentence Encoder, etc.).
  • Monetization Ready: Integrated x402 protocol for charging per tool call via Solana (devnet/mainnet).
  • Extensible: Easily add new tools via lib/ modules and tool-registry.js.
  • Ready for Agents: Full OpenAPI 3.0 specification for seamless integration with AI agents.

Setup Requirements

Prerequisites

Configuration

Copy .env.example to .env and configure the following:

VariableDescriptionDefault
PAY_TO_ADDRESSSolana wallet address to receive x402 payments.REQUIRED
FACILITATOR_URLx402 facilitator URL.https://x402.org/facilitator
X402_PRICE_USDPrice per tool call in USD.0.005
X402_NETWORKPayment network (devnet or mainnet).devnet
PUBLIC_ORIGINPublic URL for the server (needed for discovery).http://localhost:6350

Installation

npm install

Running the Server

Start the MCP Server

npm run mcp

Start all services (MCP, REST, Worker)

npm run start:all

Start the TUI

npm run tui

API Documentation

The server exposes an OpenAPI 3.0 specification at http://localhost:6350/openapi.json.

See readme.txt in the root for a summary of available endpoints.

Contributing

  1. Fork the repository.
  2. Create your feature branch.
  3. Commit your changes.
  4. Push to the branch.
  5. Open a Pull Request.

License

MIT