What to use instead of the Dome API

The product

Dome sold one key, one schema and one SDK over two venues that share nothing underneath. Its documentation index, read on 19 September 2026, still lists the surface: ten Polymarket endpoints (markets, events, trade history, order-book history, activity, market price, candlesticks, positions, wallet and wallet P&L), four for Kalshi (markets, trades, market price and order-book history), a matching-markets endpoint that paired the same sports event across both, Binance and Chainlink crypto prices, a WebSocket feed and an order router. The SDKs shipped for TypeScript and Python.

The dates are all on the vendor's own pages. The changelog puts the WebSocket launch at 9 November 2025 and the Polymarket order router into beta on 29 November 2025. Its last entry is 9 January 2026 — profile handles on the wallet endpoint, wallet lookup by handle, and a positions API for Polymarket. Nothing has been added since, which is eleven weeks of silence before anything was announced.

The announcement came on 19 February 2026, in a post on the vendor's own blog: Polymarket was acquiring Dome. The same post cancelled every paying subscription that day with no further charges, raised the free tier from 1 to 10 requests a second with immediate effect, kept existing paying customers' limits and upgraded features permanently, and offered dedicated support only through 31 March — past that date it committed to neither support nor an SLA.

The sentence that decides your week is on the documentation front page, where it still sits: all Dome APIs reach end of life on 28 April 2026, and readers are pointed at Polymarket's own APIs instead. That date has passed. On 19 September 2026 the API host answered an unauthenticated request with a 403 and a line telling you to get a key from the dashboard, and the dashboard's sign-in page still resolved — which shows that nobody has torn the DNS down, and shows nothing whatever about whether a key still returns data. A shutdown that has been announced and dated is a shutdown. Plan against the date, not against the 403.

What to use instead

Split the problem by what your code actually calls, because no single card covers all of it.

If you called markets, events, prices and candlesticks on both venues, you want another normalized feed. FinFeedAPI is the broadest: nine venues in one schema, both of Dome's among them, with historical trades, quotes and raw limit-order events served from flat files — which is the only way to replay an order book. It is billed per request with no recurring free allowance, so a poller that Dome served for nothing now has a meter on it. Adjacent covers exactly Dome's two venues and adds indices and cross-venue reference rates on a seat subscription, with Kalshi readable on its free tier.

If you called the wallet and wallet-P&L endpoints, nothing in this catalogue is a drop-in. Polyrama is the closest: Polymarket and Kalshi in one terminal, wallet P&L included, with a token-authenticated REST API, free to use. For a one-off answer rather than a dependency, the Dune Polymarket dashboards compute the same thing from the chain and can be forked and edited.

If you called the order router or the trading endpoints, go direct. The Polymarket CLOB API is the book, the prices and the order placement; the Polymarket Gamma API is the catalogue of events and markets and needs no key at all; the Kalshi API is REST, WebSocket and FIX for the other half. Two integrations instead of one, which is exactly the cost Dome was charging to remove.

If what you actually valued was the abstraction, CCXT is the same idea under an MIT licence: a unified API with a dedicated prediction namespace over seven event venues, in Python, JavaScript and six other languages. You maintain the normalisation layer instead of renting it, and nobody can sell it out from under you — which, this year, is not a theoretical advantage.

What you give up

The join between the two venues. Dome's matching-markets endpoint answered the one question that is genuinely hard: is this Kalshi ticker the same event as this Polymarket condition? Adjacent builds indices across both and FinFeedAPI makes the fields line up, but neither publishes that mapping as a maintained list. Writing it yourself is a morning; keeping it correct through a season of sports markets is a job.

Wallet P&L as a field. Polymarket's own APIs do not compute it, because it is not their data — it is a derivation over positions, fills and prices. Polyrama and the Dune dashboards give you the number; only Polyrama gives you it behind an API, and it is one small vendor's roadmap rather than a contract.

One key, one bill, one rate limit. Going direct means two auth models — a verified Kalshi account with identity checks for anything past public REST, and a Polygon wallet with a signature chain for any CLOB write — plus two id spaces, two rate-limit regimes and two things to page you at 3am. Dome's whole product was that this was one thing.

A free tier that was raised on the way out. Ten requests a second for nothing was generous, and it was generous because the service was being wound up. FinFeedAPI has no recurring free tier, and Adjacent's free tier is Kalshi only. The venues' own APIs are free and deeper, and that is the honest replacement for the free tier — at the price of doing the work Dome did.

Migration notes

Polymarket ids port; Kalshi ids do not need to. Dome addressed Polymarket markets by the same token_id the CLOB and Gamma APIs use, so that half of a rewrite is mostly renaming methods. Kalshi is ticker-addressed on both sides. The identifiers were never the hard part — the response shapes were.

The history Dome held is not being handed over. Its order-book history and candlesticks were its own storage, and no export was offered in the shutdown notice. If you need depth from before your switch-over, you are buying it again from whoever has it: FinFeedAPI's flat files, on terms whose per-venue backfill start it does not document. Check the specific venue and the specific window before committing.

Kalshi adds a KYC step Dome did not have. A Dome key needed an email. Public Kalshi market data over REST is open, but the WebSocket and every order need a verified account, which means identity documents and a jurisdiction that Kalshi serves. For a developer outside the US who was reading Kalshi through Dome, that is not a code change, it is a question of whether you get in at all.

Fail closed, not silent. Keys that stop authenticating usually surface as a 4xx your retry logic swallows. Before the next deploy, grep for the Dome hostnames in your config and make the absence of a key a startup error rather than an empty array that quietly prices a position at zero.

There is no invoice to cancel. Subscriptions were cancelled by the vendor on 19 February 2026. If a charge appears after that date, the announcement is the thing to hold it against.