Voyagier MCP server

Read from the package source without running it. The installed server may list more.

• add_traveller

  Add a traveller to a trip plan. Travellers are required before search. Gender and date of birth are required at flight checkout; passport data hard-gates international reservations (set those via the CLI travellers update later). Loyalty programs are applied at checkout best-effort — a booking never fails because of them.

• agent_docs

  Print the full Voyagier agent reference (AGENT.md) as markdown — the canonical integration guide for the compose/close loop, error codes, and quirks.

• bookwrite action

  Create a real Stripe checkout for the bookable items. REQUIRES expect_total — this is the price hard-gate: the checkout is created only if the chargeable subtotal equals expect_total exactly (cents-compared), else it fails closed with PRICE_CHANGED and NO checkout is created. Get the current subtotal from book_dry_run first. Never retry a successful book (unpaid sessions are invisible and a retry

• book_dry_runwrite action

  Preview a checkout WITHOUT creating one and WITHOUT needing a price gate: returns the chargeable subtotal, blockers, existing checkouts, and the next step. Optionally pass expect_total to also get a gate verdict (data.gate.{wouldPass,failReason}) — pre-verify without risking PRICE_CHANGED. This is the checkout truth; run it before book.

• booking_status

  Show payment + booking status for past checkouts on a plan (post-payment confirmation lookup). Booking-record amounts are raw CENTS.

• create_clientwrite action

  Create or return an existing advisor CRM client by email (idempotent upsert). A trip plan requires a clientId. Returns { client, ok, created }.

• doctor

  Self-check the CLI environment (auth, schema reachability, local state, version) before doing real work. Run this first when you hit an unfamiliar error.

• get_selection_options

  Read a selection's options. Search is ASYNC: when a search returns optionCount 0 the inventory fetch is still running — with wait=true (default) this polls with backoff until the status is terminal (READY / NO_RESULTS / AWAITING_INPUT / FETCH_ERROR), then returns the options to select from.

• plan_status

  ONE call answering 'what's left before this plan can book?'. Switch on data.readiness: BOOKED | READY_TO_BOOK | BLOCKED (act on data.blockers[]/nextSteps[]) | IN_PROGRESS (system is working — poll, don't act). book_dry_run is the checkout truth on any contradiction.

• plan_trip

  Scaffold a trip plan: creates the plan + a default goal graph (a round-trip + hotel TEMPLATE) and returns { tripPlanId, travellerIds, nextSteps }. It does NOT search or select — follow nextSteps to compose. Prune goals the brief doesn't need with the shape flags: one_way (drops the Return Flights goal), flight_only (drops the hotel goal), hotel_only (drops ALL flight goals). Omitting return alone

• quote

  Produce the advisor offer snapshot for a plan: items, chargeableTotal, and a machine-readable acceptance block { command, itemIds, expectedTotal }. quote's chargeableTotal equals the book price gate, so the acceptance total can never fail its own gate on an unchanged cart.

• search_activities

  Search bookable activities/experiences against the plan's Activity goal. Returns a compact envelope { selectionId, optionCount, topOptions[≤10] }. If optionCount is 0 the async fetch is still running — poll get_selection_options with wait, then select.

• search_flights

  Search flights against the plan's Flight goal (REUSES the goal's selection — does not create a new one). Returns a compact envelope { selectionId, optionCount, topOptions[≤10] } (round trips also return returnSelectionId). If optionCount is 0 the async fetch is still running — poll get_selection_options with wait. Round trip: pick BOTH legs; the SAME optionId appears in both legs' lists (leg-mirro

• search_hotels

  Search hotels against the plan's Hotel goal. Returns a compact envelope { selectionId, optionCount, topOptions[≤10] }. Prices are STAY TOTALS, not nightly. If optionCount is 0 the async fetch is still running — poll get_selection_options with wait, then select.

• select_option

  Choose an option on a selection by explicit selection + option id (defaults to choosing for all travellers). With wait=true (default), after the pick succeeds it polls until the pick is reflected server-side AND readiness settles, then returns a plan-status snapshot. A timed-out wait never means the pick failed. Round trip: call once per leg — the identical optionId on both legs is intended.

• Code scan63 source files scanned25/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitystatic API keys via environment variables6/15
• Maintenancelast push 0 days ago15/15
• Maintainer identitynamespace and repository owner differ; GitHub account older than a year; website matches verified namespace8/10

Safety score

83/100

higher than 68% of 4,460 AI and agents servers

Tools

15

the typical AI and agents server has 6

Tools that change things

20%

typical: 0%. Fewer write tools means less an agent can break

GitHub stars

1

more than 50% of 2,711 AI and agents servers

Downloads a week

71

more than 55% of 1,538 AI and agents servers with a package

Voyagier: common questions

Is Voyagier MCP server safe?

Mostly: it is graded B (83/100). Read the Voyagier safety report

How do I install Voyagier?

It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.

Does Voyagier need an API key?

Yes. The registry entry asks for VOYAGIER_TOKEN.

Is Voyagier maintained?

The last commit was 41 days ago (2026-09-24). The latest release is v2.14.0.

What can I use instead of Voyagier?

Servers from other publishers that do the same job: mcptoon MCP server, yutu MCP server and REA MCP server.

How does Voyagier compare with other AI and agents MCP servers?

safety score 83/100 (higher than 68% of 4,460 AI and agents servers); tools 15 (the typical AI and agents server has 6); tools that change things 20% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.