Caramelo Server
FreeNot checkedMCP server that exposes tools to query DulceGestion business data (orders, stock, customers) and retrieve documentation via RAG, enabling Claude to assist with
About
MCP server that exposes tools to query DulceGestion business data (orders, stock, customers) and retrieve documentation via RAG, enabling Claude to assist with pastry shop management.
README
Multi-channel chatbot for DulceGestion - a pastry shop management SaaS.
What it does
- Answers user questions about how to use the app (RAG-powered from documentation)
- Queries live business data (orders, stock, customers) via tool calling
- Works on Telegram, Web Widget (embeddable), and WhatsApp (Twilio)
- Exposes an MCP server for use with Claude Desktop/Code
Architecture
Channels (in/out) Brain Data
+----------------+ +----------------+ +------------------+
| Telegram |--->| |--->| RAG (docs) |
| Web Widget |--->| Bot Engine |--->| MCP Tools |
| WhatsApp |--->| |--->| Claude LLM |
+----------------+ +----------------+ +------------------+
Adapter pattern - each layer has swappable implementations:
- Channels: Telegram (telegraf), WebSocket (widget), WhatsApp (Twilio)
- LLM: Claude (Anthropic API) or hardcoded (keyword matching for testing)
- Tools: MCP registry with DulceGestion API client
Tech Stack
TypeScript, Express, telegraf, ws, @anthropic-ai/sdk, @modelcontextprotocol/sdk, Voyage AI (embeddings), vitest
Quick Start
# Install
npm install
# Configure
cp .env.example .env
# Edit .env with your keys
# Index documentation for RAG
npm run index-docs
# Run
npm run dev
Environment Variables
| Variable | Description |
|---|---|
TELEGRAM_TOKEN |
Bot token from @BotFather |
ANTHROPIC_API_KEY |
Claude API key |
VOYAGE_API_KEY |
Voyage AI key (for embeddings) |
DULCEGESTION_API_URL |
DulceGestion API endpoint |
LLM_ADAPTER |
claude or hardcoded |
PORT |
HTTP server port (default: 3002) |
SESSION_TTL_MINUTES |
Chat session timeout |
WIDGET_ALLOWED_ORIGIN |
CORS origin for widget |
Scripts
| Script | Description |
|---|---|
npm run dev |
Start with hot reload |
npm start |
Start in production |
npm test |
Run tests |
npm run index-docs |
Generate RAG embeddings from docs/flows/ |
npm run mcp |
Start MCP stdio server (for Claude Desktop/Code) |
MCP Server
Use with Claude Desktop or Claude Code:
{
"mcpServers": {
"dulcegestion": {
"command": "npx",
"args": ["tsx", "src/mcp/server.ts"],
"cwd": "/path/to/dulcechat",
"env": {
"DULCEGESTION_API_URL": "http://localhost:3001/api",
"DULCEGESTION_AUTH_TOKEN": "your-jwt-token"
}
}
}
}
Embeddable Widget
Add to any page:
<script src="https://your-server/dulcechat/widget.js" data-token="USER_JWT"></script>
Project Structure
src/
channels/ # Messaging adapters (Telegram, Web, WhatsApp)
llm/ # LLM adapters (Claude, hardcoded)
mcp/ # MCP tools and server
tools/ # Individual tool implementations
rag/ # RAG system (chunker, embeddings, retriever)
session/ # In-memory session store with TTL
engine.ts # Core orchestrator
index.ts # Entry point
docs/flows/ # User documentation for RAG indexing
widget/ # Embeddable chat widget (HTML + JS)
Adding New Tools
- Create
src/mcp/tools/your-tool.tsimplementingMcpTool - Register it in
src/index.ts - Automatically available in Caramelo, Claude Desktop, and Claude Code
Installing Caramelo Server
This server has no published package — it is built from source. Open the repository and follow its README.
▸ github.com/pablopelardas/dulcechatFAQ
Is Caramelo Server MCP free?
Yes, Caramelo Server MCP is free — one-click install via Unyly at no cost.
Does Caramelo Server need an API key?
No, Caramelo Server runs without API keys or environment variables.
Is Caramelo Server hosted or self-hosted?
A hosted option is available: Unyly runs the server in the cloud, no local setup required.
How do I install Caramelo Server in Claude Desktop, Claude Code or Cursor?
Open Caramelo Server on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.
Related MCPs
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
by modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
by xuzexin-hzCompare Caramelo Server with
Not sure what to pick?
Find your stack in 60 seconds
Author?
Embed badge for your README
Browse similar
All ai MCPs
