flightclaw/agents/traveler-profile

Traveler Profile, Preferences and Trip History (FlightClaw)

Set up a traveler profile once so FlightClaw can book flights with AI that fit the user. The profile holds the user, companions, travel preferences, cards, points balances and trip history. recommend_flights uses these preferences to rank cheap flights for the person asking.

When to use

• First use: onboard the user and their companions.
• The user states a preference ("I never fly red-eyes", "aisle seat").
• The user adds a card or a points balance.
• The user asks about past trips, or gives feedback after a trip.

Prerequisites

Connect the FlightClaw hosted MCP server: https://mcp.flightclaw.com/mcp (streamable HTTP, OAuth sign-in). Nothing to install.

• Claude Code: claude mcp add --transport http flightclaw https://mcp.flightclaw.com/mcp
• Cursor: One-click install, or add {"mcpServers":{"flightclaw":{"url":"https://mcp.flightclaw.com/mcp"}}} to ~/.cursor/mcp.json
• VS Code: One-click install, or code --add-mcp '{"name":"flightclaw","type":"http","url":"https://mcp.flightclaw.com/mcp"}'
• Claude (web / desktop): Settings > Connectors > Add custom connector > paste https://mcp.flightclaw.com/mcp
• ChatGPT: Settings > Apps & Connectors > Create (developer mode) > paste https://mcp.flightclaw.com/mcp

If the FlightClaw tools are not available, ask the user to connect the server first.

Tool flow

Who you are

• get_me: show the primary user (linked traveler, account email, home airports).
• set_me: set me_traveler (a saved traveler name), and optional account_email and home_airports (array of IATA codes).

Travelers

• save_traveler: name (short key), given_name, family_name; optional born_on, gender, title, email, phone_number, relationship, loyalty_programmes.
• list_travelers, get_traveler (by name), delete_traveler (by name).

Preferences

• get_preferences: show saved preferences and learnings.
• set_preferences: patch any subset. Fields that recommend_flights uses: cabin_rules (short-haul and long-haul cabin), preferred_airlines, avoid_airlines, depart_window, max_stops, redeye_ok, budget_sensitivity. Other preferences such as seat, baggage and meal can be saved too.

Cards and points

• save_card: id, issuer, product; optional network, region, notes. This records which cards the user holds for points awareness. It never stores card numbers.
• list_cards, delete_card (by id).
• set_points_balance: program and balance. list_points shows all balances.

Trips

• log_trip: after a booking, record route, depart_date, travelers; optional return_date, cabin, price, currency, order_id.
• list_trips, get_trip (by id).
• trips_followup: list completed trips that need feedback.
• record_trip_feedback: id, feedback and optional learnings (array). Lasting lessons fold into preferences.

Onboarding order

1. save_traveler for the user, then set_me.
2. save_traveler for each companion, and save_group if they travel together.
3. set_preferences with one or two questions per message.
4. Optional: save_card and set_points_balance.

Rules

• Never ask for or save card numbers, CVCs or passwords. save_card stores only issuer and product.
• Ask before you save personal data, and save only what the user gives.
• Confirm before delete_traveler, delete_card or other deletes.
• When the user books, confirm before create_checkout and show the price including the fee (5% + $3, minimum $12). See the book-flight skill.

Example prompts

• "Set me up: I'm Jack Culpan, home airports LHR and LGW."
• "I prefer economy on short flights and business on long-haul, and I avoid Ryanair."
• "I have an Amex Platinum and 120,000 Avios."
• "Show my past trips."
• "The Lisbon trip was great but the layover was too short. Remember that."

Learn more at flightclaw.com.