
Хтось знає, що означають ці наліпки «ВІЧ КУДЛА»?
Побачила фото цих наліпок у районі Електроніки та Заріччя. На них портрет жінки й напис «ВІЧ КУДЛА», причому їх клеять по кілька штук поруч. Хтось знає, що це за історія

Побачила фото цих наліпок у районі Електроніки та Заріччя. На них портрет жінки й напис «ВІЧ КУДЛА», причому їх клеять по кілька штук поруч. Хтось знає, що це за історія
How to connect your AI agent to real hotel search data via MCP
If you're building or testing agents and want a real-world data source to point them at: we've built an official MCP server for hotel search, live at mcp.winwin.travel.
Quick rundown of what it gives your agent:
There's also a live affiliate program if your agent starts generating bookings (4-10% commission at 5+/month).
Install guide: https://github.com/WinWin-travel/MCP-server
Docs: https://mcp.winwin.travel/docs
We're specifically trying to learn how different models handle our tool calls and nested JSON responses, so if you connect something up, we'd genuinely appreciate feedback here or via https://form.winwin.travel/mcp-feedback
Happy to answer setup questions in the comments.
We built an official MCP server for real hotel search (looking for feedback from agent builders)
Hi everyone! We're developing an MCP server that lets AI agents, Claude, ChatGPT, custom bots, search real hotel offers through natural language instead of a traditional website or API.
Tooling for good travel agents doesn't really exist yet, so we're trying to fill that gap with precise search and booking functionality through MCP.
What's live now:
Roadmap: full booking flow, Stripe payments, personalization and saved selections.
We're mainly after technical validation right now: is the output clear, can agents use our filters correctly, does our nested JSON structure cause context issues, how do different models (Claude, GPT-4o, local Llama) handle the tool calls.
Links (install guide, docs, feedback form) in the first comment, keeping the post itself link-light per sub rules. Happy to answer anything here!
hi everyone!
we're building an MCP server (model context protocol) for winwin.travel - a connector that lets AI agents (claude, chatgpt, your own bots) search real hotel offers through natural language instead of a traditional website or API.
good MCP tooling for travel barely exists yet, and our goal is to give developers precise search and booking functionality through a clean MCP interface. this subreddit is where we're testing it in the open and want your feedback.
what's already working:
smart hotel search (live now): query the real winwin.travel hotel database via MCP. connect from any MCP-compatible agent, including claude desktop, postman, cursor, and windsurf, at mcp.winwin.travel. results include more than name and price - full descriptions, room amenities, photos, star ratings, taxes, and cancellation policies. filter tools so agents can surface options intelligently, not just dump a list.
affiliate program (live now): commission-based payouts (4-10%) for agents generating 5+ successful bookings a month.
on the roadmap: full booking flow, secure payment via stripe, personalization, and saved searches.
get started:
install and access instructions: https://github.com/WinWin-travel/MCP-server
API docs: https://mcp.winwin.travel/docs
what we want your feedback on:
we're focused on technical validation right now, specifically - is our MCP useful, and is the output clear? are LLMs correctly using our filters? does the nested JSON response structure cause context loss? how do different models (claude, gpt-4o, local llama models) handle our tool calls?
drop your impressions and questions in the comments, or send structured feedback here: https://form.winwin.travel/mcp-feedback
glad to have you here - dig in and let us know what breaks.
I’m curious about local true crime cases from Madison or the surrounding area that people still remember or talk about today. Could be famous cases, unsolved mysteries, strange disappearances, murders, scandals, or anything that had a big impact on the community.
What case do you find the most interesting, disturbing, or memorable, and why?