LetsFG vs ETG API (RateHawk)

LetsFG

Hosted MCP with OAuth 2.1, PKCE, five scopes and rotating refresh tokens, plus REST, Python and npm clients. Tickets are bought by a LetsFG booking agent after the hold, 4 to 11 minutes per the docs, and bookings can pause with questions.

ETG API (RateHawk)

Hotel search and booking are documented in a public Swagger 2.0 file, an llms.txt index and a changelog with 64 dated entries in 2026. Access is by contract. Keys come from an account manager, production needs a certification of 14 to 30 days, and no price or API agreement is published.

LetsFG

1. Start on /v1/sandbox/ with no key; the booking states, questions and timings match production
2. After book_flight, poll get_flight_booking every 20 to 30 seconds and answer questions before expires_at_ms; an unanswered price change fails the booking and releases the hold
3. Honour retry_after_seconds on a 429. Over MCP the limit is 10 searches per 10 minutes and 100 a day, and early retries lengthen the pause
4. Fetch results with get_flight_results and the search_id rather than searching again, which spends allowance
5. Send the same idempotency_key when retrying a booking after a timeout
6. Treat a 410 as a retired route and read the replacement path in its body

ETG API (RateHawk)

1. Call /api/b2b/v3/overview/ first. It lists the endpoints the key may use and each one's request limit.
2. Keep sandbox IDs, keys and hashes inside api-sandbox.ratehawk.com. Sandbox region IDs are limited to 2011, 2395, 2734 and 6053839.
3. Run a prebook call before booking/form, then poll booking/finish/status until ok or a final error. Treat timeout, unknown and 5xx as still in progress.
4. Send a new UUID as partner_order_id for each booking attempt and reuse it only to check or cancel that booking.
5. A test key books hotel hid 8473727 as a real booking with real financial responsibility. Cancel it afterwards.