Tock via fpx (no MCP)

www.exploretock.com sits behind a Cloudflare managed challenge — a plain curl/Node request 403s every path (cf-mitigated: challenge, a "Just a moment..." interstitial), including /city and /{slug}. There is no server-side login form and no consumer token exchange. fpx routes the request through the user's own signed-in browser tab (the ContextMint Bridge extension), which has already cleared the challenge, so the same fetch succeeds. Browsing (metros, search, venue detail, availability) needs no Tock login at all — just an open, challenge-cleared tab; only the reservations list needs the tab signed in.

This is the same data the tock_* MCP tools return, reached with one-shot CLI calls instead of a running server. It is read-only — see "Why no booking/cancel" below.

One-time setup

npm install -g @fetchproxy/cli              # provides `fpx`
fpx profile add tock --domain exploretock.com
fpx pair -p tock                             # prints a pair code → approve in ContextMint Bridge

Requirements: the ContextMint Bridge browser extension installed (from https://github.com/nullnet-app/contextmint-bridge/releases — Chrome: load the chrome zip unpacked; Safari isn't available yet, so use Chrome for now), with an open www.exploretock.com tab (signed in, for the reservations tool only), and its Chrome Site access allowing exploretock.com. Pairing persists — after the first approval every later fpx call reuses it. ContextMint Bridge is the fetchproxy extension under its new name, same maintainer; its source is public at https://github.com/nullnet-app/contextmint-bridge — build it yourself or check a release zip against its published .sha256 file.

Two call shapes

Tock has two different backends behind the same bridge:

1. SSR pages (/city, /city/{slug}, /{slug}, /{slug}/search) — an ordinary GET returns full HTML with the entire dataset embedded as a JS object-literal assignment: window.$REDUX_STATE = {"app":{...}, "consumerPage":{...}, "calendar":{...}, ...}. It is not JSON — absent fields are the bare identifier undefined, so jq can't parse the raw response. Use the bundled references/extract-redux-slice.mjs (a faithful port of the MCP's own src/redux-state.ts) to pull one named slice out as real JSON, then pipe that to jq:

   fpx get 'https://www.exploretock.com/city' -p tock \
     | node references/extract-redux-slice.mjs app \
     | jq '.metros // . | length'
   

2. Authenticated GraphQL (/api/graphql/<OperationName>) — used only for the signed-in patron's reservations, which are lazy-loaded client-side and never appear in the SSR store. Clean JSON in and out — pipe straight to jq:

   fpx post-json 'https://www.exploretock.com/api/graphql/PatronReservationHistory?opname=PatronReservationHistory' \
     @body.json -p tock | jq '.data.purchases'
   

Ready-to-run paths/bodies for both shapes, plus the exact GraphQL document and jq recipes, are in references/requests.md.

The one rule: resolve the metro before searching

Tock's /city/{slug} search takes a metro slug (from /city's app slice, e.g. "chicago"), never a free-text city name. A venue's own page path is its domainName slug (from a search result, or the exploretock.com/{slug} URL directly) — no numeric-id resolution needed beyond that.

Exit codes (fetch verbs)

• 0 — success. For GraphQL, a 0 HTTP status can still carry an errors array in the body — check jq '.errors // empty'.
• 2 — bridge unavailable: extension not connected or pairing pending → run fpx pair -p tock, confirm an exploretock.com tab is open.
• 3 — bot wall: the tab hasn't cleared the Cloudflare challenge → open/ refresh a www.exploretock.com tab and retry.
• 4 — upstream non-2xx from Tock (e.g. a bad slug → 404).

A non-JSON 2xx body on the GraphQL path is almost always a bot-challenge or sign-in interstitial slipping through, not real data — don't blindly JSON.parse it.

Why no booking/cancel

Tock's booking/checkout/cancel flow is a stateful protobuf transaction (/api/ticket/*) authenticated by app-injected X-Tock-Authorization / X-Tock-Session / X-Tock-Fingerprint request headers set by Tock's own JS interceptor — not cookies. A bridge fetch (fpx included) only carries the tab's cookies, so replaying the protobuf calls 404s ("Unknown business identifier"). Reconstructing those headers would mean harvesting the user's live session secrets or forging an anti-bot device fingerprint — both out of scope. This skill (like the MCP) stays read-only; book/cancel on exploretock.com itself. Full detail: docs/TOCK-API.md in the tock-mcp repo.

If a booking was made outside this skill (on the site, by hand or by UI automation), treat it as confirmed only when a confirmation ID/URL/email was captured and it shows up in a PatronReservationHistory re-query afterward; otherwise report it as "attempted, unverified." The post-booking modal is not proof (a stale-cart confirm is a silent no-op that still renders it), and the reservations backend lags the Reservations tab by minutes — an immediate empty re-read proves nothing.

Notes

• fpx health -p tock shows bridge connection state when a call fails.
• Reservations/profile calls need a signed-in tab; browsing does not.
• This project is developed and maintained by AI (Claude).