mcpField
F
Flait Mcp

Flait Mcp

Remote
github.com/corteslabsxyzMaps & Travel1Updated 15w ago
Repository

About Flait Mcp

Flait Mcp evaluates flight prices using its collected route history and offers deal monitoring for travelers and booking agents. It also supports monitored auto-booking with a spending cap and a human confirmation step.

Key features

  1. 01Assess whether a flight price is good
  2. 02Find current route deals
  3. 03View daily route price history
  4. 04Set price watches
  5. 05Guarded auto-booking and status checks

Use cases

  1. 01Compare a route fare with its typical price
  2. 02Find currently lower-priced flights
  3. 03Watch a route and get notified when prices hit
  4. 04Arrange a booking with a spending cap and human approval

MCP support

The remote MCP server exposes tools for price scoring, deal search, route history, price watches, and watch status. Auto-booking is constrained by a spending cap and a human confirmation mechanism.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://flait-agent-production.up.railway.app/mcp/
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

6
ParameterTypeRequiredDescription
limitintegerNo

Details

From the project README

Flait — flight deal intelligence for nomads & their agents

Live board: https://flait-agent-production.up.railway.app

Not another flight-search wrapper. Flait watches prices 24/7 on nomad corridors (US/EU ↔ Latin America, US/EU ↔ Southeast Asia, and the hops between) and builds the price history needed to answer the question every traveler — and every booking agent — actually has: "is this price good for this route?"

GET /v1/score?origin=LAX&destination=BKK&price_usd=430
→ {"verdict":"good","pct_vs_typical":24.2,"typical_price_usd":567,"days_of_history":11,...}

History can't be scraped retroactively. Every day the collector runs, the oracle gets harder to replicate — and more useful to you.

Connect your agent (30 seconds)

Remote MCP endpoint (Claude, Hermes, OpenClaw, anything MCP):

https://flait-agent-production.up.railway.app/mcp/

Claude Desktop / Claude Code:

{"mcpServers": {"flait": {"type": "http", "url": "https://flait-agent-production.up.railway.app/mcp/"}}}

READ tools (free): is_good_deal · find_deals · route_history

ACT tools (guarded): set_watch — your agent creates a standing watch on ANY route (watching a route starts price collection on it), optionally armed to auto-book when the price hits, with the user's spend cap enforced · watch_status — poll a watch and its bookings with the key set_watch returns.

Writes are guarded. Auto-booking never fires silently: the human always gets a "booking in 30 minutes" email with a stop button first, spend is hard-capped at their number, and one watch books at most one flight. Your agent can also register a webhook_url and receive signed lifecycle events: hit / booking_scheduled / booked / booking_failed / booking_stopped. (Booking is in beta test mode — no real charges.)

Per-platform setup (Hermes config, OpenClaw, example agent missions): see AGENTS.md.

REST API

Full OpenAPI schema at /openapi.json — agent-readable by design.

EndpointWhat it answers
GET /v1/score?origin=&destination=&price_usd=Is this price good? (the oracle)
GET /v1/dealsBest current deals vs each route's typical price
GET /v1/history/{origin}/{destination}Daily price history for a route
POST /v1/intentsSet a watch (mandate fields = auto-book; webhook_url = agent callbacks)
GET /v1/watch/status?k=Watch + booking status for a status_key

Humans welcome too

The deal board shows the same intelligence visually — and you can set a standing watch ("BKK under $450") and get an email the moment it hits.

Honest notes

  • Verdicts are backed by our own collected history. Routes without ≥3 days of data return "verdict": "unknown" — we don't guess.
  • Prices come from cached airline-search data: great for deal intelligence, always verify the live fare at booking.
  • Free during beta. Be reasonable (≤1 req/sec). Building on Flait? Say hi — high-volume keys and webhooks are coming.

Built in public by one human + Claude.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.