Add real-time local restaurant data to any Langflow agent flow. Register Loop's MCP server in Settings > MCP Servers, drag the MCP Tools node onto your canvas, and wire it to an Agent node. All four Loop tools appear automatically. No API key required on the free tier.
In Langflow, open Settings > MCP Servers(or the MCP section in the flow editor's left sidebar) and click Add MCP Server. Select HTTP transport, name it Loop, and set the URL below. Save — Loop appears in the MCP sidebar.
https://stayinloop.dev/mcp# In Langflow: Settings > MCP Servers > Add MCP Server # # Transport: HTTP # Name: Loop # Server URL: https://stayinloop.dev/mcp # # With an API key (higher-volume projects): # Headers: # Authorization: Bearer <your-api-key> # # Save → Loop appears in the MCP sidebar. # Drag it onto the canvas to create the MCP Tools node.
After saving, Loop appears in the MCP sidebar. Drag it onto the canvas — Langflow creates an MCP Toolsnode pre-configured with Loop's server URL. Loop's four tools (search, get_details, verify, report) are immediately available to any Agent node you connect it to.
# Langflow agent flow — restaurant search # # Nodes: # 1. Chat Input # ↓ # 2. Agent (LLM: GPT-4o or Claude) # Tools port ← MCP Tools (Loop) # System prompt: "You help users find restaurants in Kreuzberg. # Use search() first, then get_details() when # the user picks one. Call verify() if they ask # about current hours. Call report() after they visit." # ↓ # 3. Chat Output # # Wire: MCP Tools Toolset → Agent Tools # Chat Input Output → Agent Input # Agent Response → Chat Output Input
After the user acts on a result, the agent calls report() via the MCP Tools node. The result_token comes from the prior get_details() call and expires after 30 minutes.
# After user visits the restaurant: # The agent calls report() via the MCP Tools node: { "result_token": "<token from get_details>", "outcome": "correct" // correct · wrong · booked · closed · other } # result_token expires after 30 minutes. # Each report() call updates the record's confidence score in Loop's DB.
| Tool | When to call | Key output |
|---|---|---|
search() | User asks for a restaurant | Up to 8 ranked records |
get_details() | User selects a result | Full record + result_token |
verify() | User asks about current status | Live observation + confidence |
report() | After user acts on result | Mutates record confidence |
Full schema: api.stayinloop.dev/v1/openapi.json
Go to Settings > MCP Servers and click Add MCP Server. Select HTTP transport and set the URL to https://stayinloop.dev/mcp. Save, then drag the Loop server from the MCP sidebar onto the canvas to create an MCP Tools node. Wire its Toolset port to an Agent node's Tools port.
MCP Tools is a Langflow node that connects a flow to any external MCP server over HTTP. Once you register Loop and drag it from the MCP sidebar, Langflow creates an MCP Tools node pre-configured with Loop's URL. All four Loop tools — search, get_details, verify, report — are immediately available.
The result_token is returned by get_details() and is required to call report(). It is a signed HMAC token that expires after 30 minutes. In Langflow, the agent holds it in its context window between the get_details call and the subsequent report call.
No. The free tier accepts unauthenticated connections to https://stayinloop.dev/mcp. For higher-volume projects, apply for a key at stayinloop.dev/#pricing and add it as an Authorization: Bearer <key> header when registering the server in Settings > MCP Servers.