Google Flights MCP: real-time fares your agent can search across a whole date range, ad-free

Real-time Google Flights fares for agents. Three things people do with this server. Scan for deals: one call takes a date range and a list of destination airports, expands every combination server side, and returns each fare with Google's own low, typical or high verdict. Put live search in your app: flat JSON with a bookable link on every result, and round trips priced as paired legs. Run a 24/7 AI travel agent: add the server, sign in with Google, and schedule it. No ads, no sponsored content. You bring your own RapidAPI key, so every search is billed to your plan and never to anyone else's.

One URL. Sign in with Google, nothing to paste into your client:

claude mcp add --transport http google-flights https://flights.flightpowers.com/mcp

Your client hits the URL, gets a 401 with the sign-in details, and shows a Sign in button. You sign in with Google and paste your RapidAPI key once on the /connect page. Nothing goes in your client config.

Same URL with your key, for scripts, CI and clients without a sign-in button:

claude mcp add --transport http google-flights https://flights.flightpowers.com/mcp --header "x-rapidapi-key: YOUR_RAPIDAPI_KEY"

A headless run cannot open a browser, so it sends the key instead. There is no second endpoint to switch to.

Hosted. Nothing to clone, nothing to build. Listed in the official MCP Registry as com.flightpowers/google-flights. Health check: /health.

Need a key? Subscribe to the Google Flights Live API on RapidAPI, free tier available, and copy your x-rapidapi-key: https://rapidapi.com/mtnrabi/api/google-flights-live-api

Cursor Marketplace

Once listed, install directly from the Cursor Marketplace. Until then, add the remote MCP server manually:

# In Cursor, add via Settings → MCP Servers → Add Server
# URL: https://flights.flightpowers.com/mcp
# Header: x-rapidapi-key: YOUR_RAPIDAPI_KEY

Or use the mcp.json at the root of this repo, which references ${RAPIDAPI_KEY} as a plugin variable.

What your agent gets

Two tools that answer a fare question, not a date lookup.

• Ask open-ended questions. "Cheapest one-way to Sri Lanka anywhere in October", "5 to 7 nights in Rome sometime in May, from Tel Aviv or Larnaca": each is one tool call. Both tools take a departure date range, a list of destination airports, and (round-trip) a nights value instead of a fixed return date, and expand them internally.

• Say whether a price is actually good. Every result carries Google's own historical range for that route and period: price_insights_low, price_insights_high, and a price_range_in_relation_to_other_periods verdict of low / typical / high. That is what lets an agent answer "$209 is typical here, don't rush" instead of just quoting a number.

• Book, not just browse. Every result includes a buy_link to Google Flights.

• Know what it spent. Every response carries api_usage: requests used by this call, and what is left on the caller's plan. See Spend reporting.

• Know what it searched. Every response carries search_coverage, so the model can say honestly which dates and destinations the answer is based on.

Results are live fares. They go stale within minutes: never cache a fare or reuse an earlier result; search again and state when the data was fetched.

Get a key (free tier available)

The server holds no upstream credential of its own. Every search is billed to your RapidAPI subscription, which is why the key travels with the request.

1. Subscribe to the Google Flights Live API: https://rapidapi.com/mtnrabi/api/google-flights-live-api

2. Copy your x-rapidapi-key.

3. Pass it to the server in any one of the three ways below.

If a key is missing, the tools do not fail silently and do not spend anything. They return needs_api_key: true with the signup URL and these instructions, phrased for the model to read back to you.

Three ways to pass your key

First non-empty source wins, in that order. The key is never logged, never echoed into an error message, and never returned in a tool response.

Gemini CLI

Install via the Gemini extensions CLI:

gemini extensions install https://github.com/mtnrabi/google-flights-mcp

The installer will prompt for your RapidAPI key. Subscribe at https://rapidapi.com/mtnrabi/api/google-flights-live-api (free tier available) and copy your x-rapidapi-key.

Sign in once at /connect

This is the page the sign-in URL at the top of this README sends you to. A client that speaks MCP authorization walks you through it on its own; the steps below are the same thing done by hand.

Where a deployment has it enabled (check connect_enabled on /health), there is a page at /connect that replaces all of the above with a sign-in:

1. Open https://flights.flightpowers.com/connect (hotels: https://hotels.flightpowers.com/connect) and sign in with Google.

2. Paste your RapidAPI key once, into a form, over TLS.

3. Press Reveal the URL and copy the connect URL, …/mcp?fp_token=fpk_…, and use that as the server URL in your MCP client. Clients that let you set headers can send the same token as Authorization: Bearer fpk_… instead. (It is hidden until you ask for it: that URL is a 90-day bearer credential for your plan, and a page that prints one by default puts it in every screenshot and screen share.)

If your client signed you in itself -- Claude, Cursor, ChatGPT and anything else that speaks MCP authorization -- there is no URL to copy. /connect says so: it shows the key you connected and tells you to go back to your assistant. There is a link on it for the case where a second client cannot sign in and does need a connect URL.

What that buys you: your RapidAPI key is not in your client config, not in a URL, and not in whatever logs that URL passes through. What it costs: the server stores your key, encrypted, and knows your Google account id and email address. Disconnect on the same page deletes the record and kills every connect token for your account, immediately. The full description is section 2a of the privacy policy.

Some details worth knowing:

• Saving runs one check. The key is validated against the listing before it is stored, so a typo fails on the page rather than in your client an hour later. That check costs at most one request from your own plan: on the free BASIC plan (10 a month), one of ten. A key that RapidAPI rejects at the gateway costs nothing.

• A key on the request always wins. If you send an x-rapidapi-key header (or any of the other channels above) and carry a connect token, the request's own key is used. Nothing you already have set up changes behaviour because you signed in.

• The token is not your key and cannot be turned back into it. It is valid for 90 days, and it stops resolving the moment you disconnect. A call carrying a token whose key has been disconnected gets a needs_api_key reply telling you to reconnect. It never falls back to somebody else's subscription and never spends anything.

• BASIC is free. Google Flights Live API · Booking Live API. One RapidAPI key covers whichever of the two you have subscribed to; you connect it once.

Running /connect on your own deployment

Off unless all four of these are set. A half-configured deployment registers none of the routes and serves keyed callers exactly as before; /health reports connect_enabled so that is visible rather than guessed.

Optional: MCP_CONNECT_VALIDATE=0 stores a pasted key without checking it first.

Authorised redirect URIs to register on the Google client, one per product origin, exactly:

https://flights.flightpowers.com/connect/callback
https://hotels.flightpowers.com/connect/callback

The alias google-flights-mcp.flightpowers.com needs no entry. /connect and /connect/start bounce an alias to the canonical origin before the sign-in starts, because cookies are per-host and Google compares redirect_uri literally: an alias that started its own sign-in would come back to a host with no state cookie and fail with a message that reads like a Google misconfiguration.

Also on the OAuth consent screen: scopes openid and .../auth/userinfo.email, and nothing else.

Rotating MCP_KEY_MASTER logs everybody out and invalidates every stored key. That is deliberate: after a rotation nothing is left holding a token that resolves to a key nobody can read. Users see "connect again", not a failed search. key_version on the table is there so a staged rotation is possible later without a flag day.

Verifying the flow end to end

Migration first, once per database:

psql "$DATABASE_URL" -f migrations/001_mcp_user_keys.sql

Then, after deploying:

# 1. The feature is actually on.
curl -s https://flights.flightpowers.com/health | grep connect_enabled

# 2. The page renders for an anonymous visitor.
curl -sI https://flights.flightpowers.com/connect        # 200
curl -sI https://flights.flightpowers.com/connect/start  # 302 to accounts.google.com

# 3. Sign in in a browser, paste a key, press Reveal and copy the connect URL.

# 4. MCP Inspector against that URL -- list the tools, then run one real search.
npx @modelcontextprotocol/inspector
#   Transport: Streamable HTTP
#   URL: https://flights.flightpowers.com/mcp?fp_token=fpk_...

# 5. Claude Code, the same URL.
claude mcp add --transport http flightpowers \
  "https://flights.flightpowers.com/mcp?fp_token=fpk_..."
claude mcp list          # shows it connected
#   then, in a session: ask for a fare and check the result is real

# 6. Cursor: Settings -> MCP -> Add, same URL. Or in ~/.cursor/mcp.json:
#   { "mcpServers": { "flightpowers": {
#       "url": "https://flights.flightpowers.com/mcp?fp_token=fpk_..." } } }

# 7. Hotels, the other hostname, with the same token.
#   https://hotels.flightpowers.com/mcp?fp_token=fpk_...

# 8. Press Disconnect on /connect, then re-run step 4. The tool must answer
#    needs_api_key with a "connect again" message -- not a search, and not a
#    generic "get a key" reply.

A tools/list that succeeds proves nothing about any of this: a token is only consulted when a tool actually runs. Step 4 has to be a real search.

Sign in from inside your MCP client

/connect works by hand, but a client only starts a sign-in on its own when a request comes back 401 with a WWW-Authenticate: Bearer resource_metadata=… header. /mcp does exactly that now, and only when it has to:

https://hotels.flightpowers.com/mcp behaves the same way for hotels. One URL per product, whichever way you authenticate, and there is nothing to choose between when you add the server.

The older …/mcp/oauth address still answers as the always-challenge alias, for connectors saved on it before 2026-09-09; nothing new needs it.

What it is like to use. Paste the URL into your client. It registers itself, opens a browser, you sign in with Google, and you approve that client by name on a page that says exactly what it will be able to do: run searches billed to your own RapidAPI plan, nothing else. The client never sees your RapidAPI key. If you have not connected one yet, the approval still works and the first search comes back telling you to paste a key at /connect, with the URL.

What you can revoke, and how. Press Disconnect on /connect: the stored key is deleted and every OAuth token for that Google account is dropped in the same action. A client that was connected stops working immediately. Individually, a client can call /oauth/revoke (RFC 7009).

Client setup

# Claude Code
claude mcp add --transport http flightpowers \
  "https://flights.flightpowers.com/mcp"
claude mcp list            # shows "needs authentication" until you sign in
/mcp                       # in a session: pick the server, follow the sign-in

# Cursor -- Settings -> MCP -> Add, URL above. Or ~/.cursor/mcp.json:
#   { "mcpServers": { "flightpowers": {
#       "url": "https://flights.flightpowers.com/mcp" } } }
# Cursor discovers the 401, registers itself and opens the browser.

# ChatGPT -- Settings -> Connectors -> Create. It asks for:
#   MCP server URL:  https://flights.flightpowers.com/mcp
#   Authentication:  OAuth
# Leave client id and secret EMPTY: this server supports dynamic client
# registration, so ChatGPT registers itself. Nothing else has to be filled in.

# MCP Inspector -- the quickest way to watch the whole handshake.
npx @modelcontextprotocol/inspector
#   Transport: Streamable HTTP
#   URL: https://flights.flightpowers.com/mcp
#   Auth: OAuth 2.0  ->  "Guided OAuth Flow" walks metadata -> DCR ->
#   authorize -> token, and shows each response. Then run ONE real search.

The protocol surface

The authorization endpoint is under /connect on purpose: the sign-in session cookie is scoped Path=/connect so it can never be attached to a /mcp request, and putting authorize anywhere else would mean either widening that cookie or making the user sign in twice.

Codes live 10 minutes and are single-use (DELETE … RETURNING, so two concurrent exchanges race on one row and exactly one wins). Access tokens live 1 hour, refresh tokens 30 days with rotation. Everything is opaque and stored as a SHA-256 hash, so a dump of the database contains nothing that can be replayed. Tokens are not JWTs, deliberately: a signed token stays valid until it expires whatever we decide afterwards, and Disconnect has to mean disconnect.

An access token is checked against the resource it was approved for before it is accepted, not only when it is issued. Both products are the same deployment, the same database and the same stored RapidAPI key per user, so without that check a token approved on the flights consent page, which says "search live flight fares" and nothing else, would be accepted on the hotels hostname and spend the user's hotels plan. The check is strict about the host and forgiving about the path, because clients in the wild send the origin, /mcp and /mcp/oauth for the same server. tests/test_oauth.py::TestATokenIsBoundToTheResourceItWasApprovedFor pins both halves.

Running it on your own deployment

No new environment variable. It comes on wherever /connect is configured, because it reuses that Google sign-in and that key store. It needs one more table in the same database:

psql "$DATABASE_URL" -f migrations/002_mcp_oauth.sql

/health then reports oauth_enabled: true and oauth_mcp_endpoint; that URL is what goes in a directory listing. MCP_OAUTH=off disables it while leaving /connect running; that is the rollback that needs no code change.

Nothing has to change on the Google OAuth client. The redirect URI is still …/connect/callback, because the MCP client's OAuth flow ends at our authorize page, and only that page talks to Google.

Verifying it end to end

BASE=https://flights.flightpowers.com

# 1. On, and advertising itself.
curl -s $BASE/health | python3 -m json.tool | grep oauth_

# 2. The challenge, on the one URL, with nothing presented. This is the
#    whole feature in one response.
curl -si -X POST $BASE/mcp -H 'content-type: application/json' \
  -H 'accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"1"}}}' | head -20
#   HTTP/2 401
#   www-authenticate: Bearer resource_metadata="…/.well-known/oauth-protected-resource/mcp"

# 3. Discovery, at the scoped and the bare path.
curl -s $BASE/.well-known/oauth-protected-resource/mcp | python3 -m json.tool
curl -s $BASE/.well-known/oauth-authorization-server | python3 -m json.tool

# 4. Dynamic registration answers.
curl -s -X POST $BASE/oauth/register -H 'content-type: application/json' \
  -d '{"client_name":"probe","redirect_uris":["http://127.0.0.1:9999/cb"],
       "token_endpoint_auth_method":"none"}' | python3 -m json.tool

# 5. The real test: MCP Inspector, Guided OAuth Flow, then ONE real search.
#    A tools/list proves nothing -- the token is only consulted when a tool runs.

# 6. Hotels, the other hostname, same walk: https://hotels.flightpowers.com/mcp

# 7. A keyed call on the same URL. This must still work, with no 401 anywhere:
curl -s -X POST $BASE/mcp -H 'content-type: application/json' \
  -H 'accept: application/json, text/event-stream' \
  -H "x-rapidapi-key: $REAL_KEY" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"1"}}}'

# 8. Disconnect on /connect, then re-run step 5's search: it must answer
#    needs_api_key, and the client must be logged out.

Keeping the registration table honest

/oauth/register is open, because the MCP spec requires it and because a client_id on its own authorises nothing: every flow through one still ends at a consent page a signed-in human has to press a button on. Open is not the same as unlimited, so three things stand behind it.

The rate limit is per INSTANCE. On Vercel that means N warm instances allow up to N times those numbers between them, and a cold start starts the counters at zero. That is why the durable caps exist as well: they are counted in the database, where the number is the same everywhere.

The address every one of these is keyed on comes from x-real-ip first -- Vercel sets it to the peer it accepted -- and otherwise from the LAST usable entry of x-forwarded-for, skipping hops that can only be internal. Never entry 0: proxies append on the right, so the left-most entry is whatever the caller wrote, and reading it would make every limit here one header away from being bypassed. A request with neither header shares one bucket named unknown, which is rate limited as a single caller and is not subject to the durable per-address cap (one missing header on the edge must not lock the whole server out for a day).

A registration is stamped as in-use when its consent page is RENDERED, not only when the human presses Approve. MCP clients commonly register when they are installed and authorize days later, and the sweep must not delete a row while its consent page is on screen -- there is no foreign key from codes or tokens back to the client, so the exchange that followed would fail invalid_client with nothing naming the cause.

Requires one migration:

psql "$DATABASE_URL" -f migrations/003_mcp_oauth_hygiene.sql

Refresh tokens rotate, and a replay revokes the family

A refresh token is single-use: exchanging it issues a new pair and retires the one presented. The retired row is kept and stamped, not deleted, because a deleted row and a token that was never issued look identical -- and telling those apart is the point. Presenting an already-rotated refresh token means either a client that lost the response or a copy in somebody else's hands, and OAuth 2.1 §4.14.2 says to assume the second: the answer is invalid_grant, and every token descended from that authorization is deleted. The honest client signs in again; the thief's access token stops working at the same moment.

With one deliberate exception, for the case that is almost always the innocent one: the FIRST replay of the token we just rotated, from the same client, within 10 seconds, is answered with the pair that rotation already issued. It is an idempotent retry -- nothing new is created -- and it means a client whose response was lost to a dropped connection is not silently signed out. A second replay, or one after the window, is the real thing and still kills the family. The window is per instance and in memory, so a miss simply falls through to the conservative answer.

Revoking a refresh token through /oauth/revoke takes its access tokens with it, for the same reason (RFC 7009 §2.1) -- and only if the token was issued to the client asking, which is the other half of that section. A client presenting somebody else's token still gets 200 (§2.2) and nothing is revoked.

A client_id can be a URL

client_id_metadata_document_supported: true is advertised in /.well-known/oauth-authorization-server. A client may use an https URL as its client_id; the document at that URL lists its redirect_uris, and we fetch and check it per flow instead of writing a registration row. Smithery asks for this before it will proxy a remote OAuth server.

What is checked, every time: https only, a public hostname (no IP literals, no localhost, no credentials in the URL), the hostname resolved and every address it answers with required to be public unicast (a name is not a control: 127.0.0.1.nip.io has a dot in it and points at loopback), no redirect followed, a 5-second timeout, a 64 KB cap enforced while the body is read rather than after it is buffered, a client_id inside the document that matches the URL if it is present, and -- the one that matters -- the redirect_uri in the request must be listed in the document. The lookup is rate limited on its own (60 per address per 10 minutes), because it is the only outbound fetch in this server that a caller can trigger before signing in. Dynamic registration is unchanged and still the default: a client_id that is not an https URL is looked up in the table exactly as before.

Tools

The hotels deployment serves search_hotels, find_hotel_by_name and compare_hotel_rates instead; see Hotels: providers and compare_hotel_rates.

search_oneway_flights

search_oneway_flights(
    from_airport: str,                     # origin IATA, e.g. "TLV"
    to_airport: str | list[str],           # destination IATA, or a list to compare
    departure_date: str | None = None,     # "YYYY-MM-DD"
    departure_date_from: str | None = None,# first date of a range
    departure_date_to: str | None = None,  # last date of a range
    max_stops: int | None = None,          # 0 = non-stop only
    airline_codes: list[str] | None = None,
    exclude_airline_codes: list[str] | None = None,
    departure_time_min: int | None = None, # hour, 0-23
    departure_time_max: int | None = None,
    arrival_time_min: int | None = None,
    arrival_time_max: int | None = None,
    currency: str = "usd",
    max_price: int | None = None,
    seat_type: int | None = None,          # 1 economy, 2 premium economy, 3 business, 4 first
    passengers: list[int] | None = None,   # [adults, children, infants]
    sort_by: str = "best",                 # "best" | "price" | "duration"
    limit: int = 10,                       # results returned after merge + sort
    max_searches: int | None = None,       # cap the billed requests this call may make
    use_fallback: bool | None = None,      # leave unset: accepted upstream, currently inert
)

search_roundtrip_flights

search_roundtrip_flights(
    from_airport: str,
    to_airport: str | list[str],
    departure_date: str | None = None,
    departure_date_from: str | None = None,
    departure_date_to: str | None = None,
    return_date: str | None = None,        # use this OR nights, not both
    nights: int | list[int] | None = None, # e.g. 7, or [5, 6, 7]
    max_departure_stops: int | None = None,
    max_return_stops: int | None = None,
    departure_airline_codes: list[str] | None = None,
    return_airline_codes: list[str] | None = None,
    currency: str = "usd",
    max_price: int | None = None,
    seat_type: int | None = None,
    passengers: list[int] | None = None,
    sort_by: str = "best",
    limit: int = 10,
    max_searches: int | None = None,
    use_fallback: bool | None = None,
)

sort_by is applied by this server across the merged result set from every search it ran, so it is predictable regardless of how many combinations were expanded.

A worked example

User: "I'm in Tel Aviv. Cheapest week-long trip to Rome or Athens, leaving any day in the first half of May."

One call:

{
  "name": "search_roundtrip_flights",
  "arguments": {
    "from_airport": "TLV",
    "to_airport": ["FCO", "ATH"],
    "departure_date_from": "2026-05-01",
    "departure_date_to": "2026-05-15",
    "nights": 7,
    "sort_by": "price",
    "limit": 5
  }
}

That expands to 15 dates × 2 destinations = 30 combinations, which is exactly the per-call cap. The response shape (field names are real; the values below are illustrative, not a quote, run the call to get live fares):

{
  "results": [
    {
      "from_airport": "Tel Aviv (TLV)",
      "to_airport": "Rome (FCO)",
      "departure_date": "2026-05-05",
      "return_date": "2026-05-12",
      "total_price": "$XXX",
      "total_price_as_number": 0,
      "total_duration_seconds": 0,
      "total_stops": 0,
      "price_range_in_relation_to_other_periods": "low",
      "price_insights_low": 0,
      "price_insights_high": 0,
      "departure_flight_airline": "...",
      "departure_flight_departure_description": "...",
      "departure_flight_arrival_description": "...",
      "departure_flight_duration": "...",
      "departure_flight_stops": 0,
      "departure_stops_info": [],
      "return_flight_airline": "...",
      "return_flight_departure_description": "...",
      "return_flight_arrival_description": "...",
      "return_flight_duration": "...",
      "return_flight_stops": 0,
      "return_stops_info": [],
      "buy_link": "https://www.google.com/travel/flights?tfs=..."
    }
  ],
  "result_count": 5,
  "search_coverage": {
    "requested_combinations": 30,
    "searched_combinations": 30,
    "truncated": false,
    "max_searches_per_request": 30,
    "departure_dates_searched": ["2026-05-01", "..."],
    "destinations_searched": ["ATH", "FCO"]
  },
  "api_usage": {
    "requests_used_by_this_call": 30,
    "plan_requests_remaining": 0,
    "plan_requests_limit": 0,
    "note": "This search used 30 of your RapidAPI plan's requests; ... remain in the current period. Each date and destination combination is one billed request."
  }
}

Other response shapes to expect, all of them normal:

• No flights on those dates. results: [] with a message: Google Flights genuinely returns nothing for some route/date combinations. Not an error. Try nearby dates or a nearby airport. use_fallback will not change this and is left unset by default: the backend accepts the field, but the second flight-data source it selects is gated behind USE_FALLBACK_FLI (fallback_available()), which is not switched on for this API, so none of its three values has any observable effect on a search today. The automatic retries the backend does on an unreadable page are unconditional and are not affected by it.

• Some searches failed. A partial field says how many of the executed searches failed, and the results cover the rest.

• Range too wide. search_coverage.truncated: true plus a note. The range is sampled evenly across the whole window (first and last kept), not cut short, so the sample is representative, not the first N days. Raise max_searches or narrow the range for fuller coverage.

• No key / rejected key. needs_api_key: true, zero spend, with the fix. A valid RapidAPI key that is not subscribed to this API is the most common cause.

• Plan exhausted. quota_exhausted: true with api_usage, plus a reminder that narrowing the range makes remaining quota go further.

Hotels: a check-in range

POST /search prices exactly one stay, so "cheapest three nights in Rome in May" used to be 31 tool calls -- or, in practice, one call on a date the model picked and an answer presented as the cheapest. Both hotel search tools now take the flights shape instead:

search_hotels(
    destination: str,
    checkin_date: str | None = None,        # one stay: this plus checkout_date
    checkout_date: str | None = None,
    checkin_date_from: str | None = None,   # or a range: this, checkin_date_to and nights
    checkin_date_to: str | None = None,
    nights: int | list[int] | None = None,  # 3, or [2, 3, 7] to price several lengths
    max_searches: int | None = None,        # cap the billed requests this call may make
    ...
)

Same machinery as the flights fan-out (src/fanout.py): one backend call per stay, capped at max_searches_per_tool_call (30, hard max 60), sampled evenly across the range when it does not fit, and reported in search_coverage. nights derives each check-out date, so it replaces checkout_date rather than joining it. A fixed checkout_date against a range of check-in dates is allowed and means "out on the 4th, whenever I arrive"; the impossible pairs are dropped.

The response is bounded on purpose. Every property of every stay is ~25 KB per stay (measured: 25,892 bytes for 25 properties, 18,989 of them URLs), so each stay reports its cheapest property, its per-night rate and its median, and the full property list comes back for the cheapest stay only:

{
  "results": [ /* every property of the CHEAPEST stay, upstream rows untouched */ ],
  "result_count": 18,
  "results_for_stay": {"checkin_date": "2026-05-12", "checkout_date": "2026-05-15", "nights": 3},
  "stays": [
    {
      "checkin_date": "2026-05-01", "checkout_date": "2026-05-04", "nights": 3,
      "search_status": "ok", "reason": "ok",
      "property_count": 22, "priced_count": 19,
      "cheapest_total": 411.0, "price_per_night": 137.0, "median_total": 690.0,
      "currency": "USD",
      "cheapest": { /* the row, minus its image URL */ }
    },
    {"checkin_date": "2026-05-02", "search_status": "degraded", "reason": "search_failed",
     "property_count": null, "priced_count": null, "cheapest": null},
    {"checkin_date": "2026-05-03", "search_status": "not_searched", "reason": "not_searched",
     "property_count": null, "cheapest": null}
  ],
  "cheapest_overall": {"checkin_date": "2026-05-12", "total": 305.0, "price_per_night": 101.67,
                       "currency": "USD", "property": { /* ... */ }},
  "search_status": "partial",
  "search_coverage": {
    "requested_combinations": 31,
    "searched_combinations": 15,
    "truncated": true,
    "max_searches_per_request": 30,
    "stays_searched": [{"checkin_date": "2026-05-01", "checkout_date": "2026-05-04"}, "..."],
    "checkin_dates_searched": ["2026-05-01", "..."],
    "note": "This request expanded to 31 stays, above the ..."
  },
  "api_usage": {"requests_used_by_this_call": 15, "note": "... Each stay -- one check-in date paired with one length -- is one billed request."}
}

reason on a stay is a fact about our pipeline, never a guess about the hotel:

Counts are null rather than 0 on the last two: zero reads as "nothing there", and neither case knows that. Top-level search_status is ok / partial / empty / degraded over the stays; every stay failing raises instead of answering with an empty list.

Two deliberate refusals: a check-in range with providers naming more than one source (a fan-out times a per-source fan-out, billed to two subscriptions, that neither search_coverage nor api_usage can describe honestly today), and max_searches on a single stay, which would silently do nothing.

A single stay is byte-identical to what it was before this existed -- same request body, same response keys, no stays, no search_coverage, no search_status. Asserted in tests/test_hotel_date_range.py::TestTheOldShapeIsUntouched against a frozen expectation captured from the previous code.

Hotels: providers and compare_hotel_rates

The hotels deployment (hotels.flightpowers.com, the same code selected by the Host header) serves three tools. search_hotels and find_hotel_by_name also take a check-in range (above); search_hotels gained one optional argument for sources and there is one new tool.

The default did not move

search_hotels with no providers argument sends the same upstream request it always sent, to the same host, and answers with the same keys. So does providers: ["booking"]. Both are asserted in tests/test_providers.py::TestTheDefaultDidNotMove, upstream request body included -- a default is only a default if keeping it costs nothing.

Naming a second source changes the response shape, and only then:

{
  "results": [ /* every source's rows, each carrying "provider" and "rating_scale" */ ],
  "result_count": 3,
  "providers":        [ /* one row per source that was CALLED */ ],
  "providers_skipped":[ /* one row per source that was NOT, with a subscribe_url */ ],
  "caveats":          [ /* what to read before calling one source cheaper */ ],
  "api_usage":        { "requests_used_by_this_call": 2 }
}

Where each source is called

• booking goes straight to booking-live-api.p.rapidapi.com on the caller's key, billed by RapidAPI to their own subscription. Unchanged.

• airbnb goes through our own front door, POST https://api.flightpowers.com/v1/hotels/search with provider: "airbnb" in the body, because the Airbnb backend is not on the RapidAPI edge. The call carries the caller's key as x-rapidapi-key and identifies itself with X-FP-Client: mcp-hotels/<version> -- the front overwrites body attribution with its own conclusion, so the header is the only thing that says who called (rule 11). Override the origin with API_FRONT_BASE_URL for a preview front; it holds no credential.

Airbnb is available when the Airbnb listing launches on RapidAPI. It does not exist there yet, and the front ships with the provider off, so today a real providers: ["airbnb"] call comes back as a degraded row saying so. That is the designed answer, not a bug: an unmetered source reachable by anyone with any valid key is a gateway we pay for.

Four rules the implementation holds

1. A source you have no key for is never called on ours. It is named in providers_skipped with reason (no_key, not_subscribed, key_rejected) and the URL where you subscribe. Each listing is a separate subscription, so a 403 from the Hub means "not subscribed to that listing", not "bad key".

2. A degraded source is a named row, not a hole. search_status: "degraded", count: null, no prices -- and the other source's rows still come back. "Booking did not answer" and "Booking had nothing" are opposite answers and must be sayable as different sentences.

3. cheapest_total and median_total cover only the rows that carry a price, and count is that number. A price_string is never parsed into a number.

4. No cross-currency arithmetic. Every source is asked for the same currency; if two answer in different ones, each row keeps its own and caveats says the totals are not comparable. Nothing is converted.

Keys, per source

Almost everybody has one RapidAPI key subscribed to several listings, and that key is used for every source with no extra configuration. A caller who genuinely holds two can name one per source, and the source-scoped name wins:

x-rapidapi-key-airbnb: <key>        # header
?rapidapi_key_airbnb=<key>          # query
?config=<base64 {"rapidApiKeyAirbnb": "<key>"}>   # Smithery blob

The order is the one src/credentials.py already documents -- source-scoped names, then the unscoped specific names, then the config blob, then generic names last and skipped entirely when a config parameter is present, because a gateway's own key under a generic name is not ours.

filters and price_as_seen_from are Booking-only

On a mixed search they are sent to Booking and not to the front. On an Airbnb-only search they are refused, not dropped: a silently discarded filter returns more properties than you asked for and nothing says so.

Structured output (outputSchema, structuredContent, isError)

Every tool declares an outputSchema, and every result carries the payload twice: once as structuredContent, once as the serialized JSON in a text content block. The MCP spec asks for the duplicate --

For backwards compatibility, a tool that returns structured content SHOULD also return the serialized JSON in a TextContent block.

-- and it is load-bearing here rather than ceremonial, because clients that predate structured output read the text block and nothing else. (Verified against spec revision 2026-07-28; structured output arrived in 2025-06-18.)

The schemas are deliberately additionalProperties: true with only results required. The spec puts the obligation on the server -- "Servers MUST provide structured results that conform to this schema" -- and these tools have several legitimate exits that carry different keys -- a zero-result answer, the keyless needs_api_key reply and the quota_exhausted reply. A tighter schema would look better and would make the server non-conformant on a path it ships on purpose.

search_status, and why degraded is an error

Flight results carry search_status, mirroring the backend's own X-Search-Status vocabulary:

A degraded result is also flagged isError: true. It is the only one that is. The spec classifies "API failures" as tool execution errors and says clients "SHOULD provide tool execution errors to language models to enable self-correction", while nothing in the spec obliges a host to show structuredContent to the model at all. A failure carried only by a field inside the payload is therefore a failure the model may never see, which was the whole problem search_status was added to solve.

The payload still rides along with the error -- structuredContent and the text block are both present, so nothing is lost. api_usage in particular: a degraded search still spent the caller's own RapidAPI requests, and hiding that would hide a charge they have to pay. empty and partial are not errors: one is a true negative and the other carries results a caller can use.

Spend reporting (api_usage)

The money is yours, so the meter is visible. Every successful response carries:

plan_requests_remaining and plan_requests_limit come from the upstream response and are omitted when upstream does not report them; the note adapts. The rule the model should state out loud: one date × one destination = one billed request.

Cost control knobs, in order of bluntness: max_searches per call (lower it to spend less on a wide question), a narrower date range, a shorter destination list.

One call vs thirty

The underlying REST API takes exactly one (origin, destination, date) tuple per call. Against a one-date-per-call passthrough, "cheapest to Sri Lanka anywhere in October" is 31 separate tool calls: 31 round trips through the model, 31 chances to lose the thread, and a bill the user only discovers afterwards.

Here it is one tool call. The fan-out happens server-side, concurrently, capped, evenly sampled, deduplicated on buy_link, merged, sorted by your sort_by, and reported honestly in search_coverage and api_usage.

This server carries no ads at all, not by taste but by constraint: Anthropic's connector directory policy and OpenAI's app guidelines both prohibit advertising and sponsored content in tool results.

Local development

git clone <this repo> && cd mcp_server_paid
python3 -m venv .venv && .venv/bin/pip install -r requirements.txt
cp example.env .env          # fill it in; leave RAPIDAPI_KEY empty
set -a && . .env && set +a
.venv/bin/python -m src      # streamable HTTP on http://localhost:8000/mcp

Point a client at the local process the same way:

claude mcp add --transport http google-flights-local http://localhost:8000/mcp --header "x-rapidapi-key: YOUR_RAPIDAPI_KEY"

Tests (791 passing, verified):

.venv/bin/python -m pytest -q

Configuration lives in example.env; every variable is documented there. The ones that matter:

Operational routes: GET /health (public, unauthenticated: registries poll it), GET /metrics, GET /metrics/calls?hours=24, GET /.well-known/mcp/server-card.json (see below).

The static server card

GET /.well-known/mcp/server-card.json returns this deployment's metadata as a standalone JSON document: serverInfo, description, transport, capabilities, authentication, instructions, and the full tools and prompts lists exactly as tools/list serialises them. Public, unauthenticated, Cache-Control: public, max-age=3600, open CORS.

It exists because /mcp answers 401 to a scanner that brings no credential, so an automated scanner cannot read the tool list off the wire. Smithery's publish page names this document as the way out: "If automatic scanning can't complete (auth wall, required configuration, or other issues), you can provide server metadata manually via a static server card at /.well-known/mcp/server-card.json". The field list follows SEP-1649.

Two things worth knowing:

• The tool list is read from the LIVE registry at first request, never written out by hand, so it cannot drift from what tools/list returns. tests/test_server_card.py compares the two on both products.

• It is per hostname, like everything else here: hotels.flightpowers.com returns the hotel card, both flights hostnames return the flights card, and every URL inside is on that product's own origin. transport.endpoint is /mcp, the single endpoint, with the always-challenge /mcp/oauth alias listed under _meta.

curl -s https://flights.flightpowers.com/.well-known/mcp/server-card.json | python3 -m json.tool
curl -s https://hotels.flightpowers.com/.well-known/mcp/server-card.json  | python3 -m json.tool

Deployment target is Vercel via api/index.py (FastAPI wrapper handing FastMCP its lifespan, stateless_http=True). The canonical MCP path is /mcp, no trailing slash.

Never commit a real key. example.env ships with placeholders; keep it that way.

Run it in a container

The hosted server needs nothing installed. This is for self-hosting, and it is what lets Glama run its build test and cut a release.

docker build -t flightpowers-mcp .
docker run --rm -p 8000:8000 flightpowers-mcp
curl http://localhost:8000/health

The container serves streamable HTTP on ${PORT}/mcp, the same transport as the hosted deployment, via python -m src. api/index.py is the Vercel wrapper and is not used here.

No secret is baked into the image. Every search is billed to the caller's own RapidAPI subscription and their key travels with the request as x-rapidapi-key. RAPIDAPI_KEY is optional and is a server-side fallback: when set, a caller who sends no key of their own is served on, and billed to, that subscription. Leave it unset unless that is what you want; /health reports server_side_key_configured either way.

# optional, local development only
docker run --rm -p 8000:8000 -e RAPIDAPI_KEY=your_key flightpowers-mcp

Every variable in the table above works as -e NAME=value. HOST defaults to 0.0.0.0 and PORT to 8000 inside the image. The HEALTHCHECK polls /health on $PORT, so overriding PORT keeps working.

Non-affiliation

This is an independent API that returns publicly available flight pricing. It is not affiliated with, endorsed by, or sponsored by Google. "Google Flights" is used only to describe the public data source. Fares are supplied by the upstream provider, change constantly, and are not guaranteed. Always confirm the price on the airline or booking site before purchase.