MCP HeartbeatTHE MCP OBSERVATORY
Public workspace

Explore

Evidence for your next connection.

Independent measurements of remote MCP servers. Every observation has a timestamp.

Read-only probes

MCP server · AI / LLM tooling

Arena Predictions

Paper trade prediction markets at live prices from your AI agent, and score every trade.

Checks passed
12 / 22

2 failed · 8 not measured

Median response
676 ms

Median of successful probes over 7 days

30-day uptime
Not measured

Share of countable probes over 30 days, including daily rollups. Auth challenges are excluded.

Discovered tools
30

Tool definitions returned by this server

Independent monitoring

Health & checks

Status history

Outdated check

Last connection measurement is over 24 hours old. Current availability is unknown.

Last checked
Check summary by test group
Test areaPassedFailedUnmeasured
ConnectionCan a client connect and discover tools?
500
Reliability & speedDoes it respond consistently and quickly?
004
Access & authenticationWhat is required to use this server?
102
Tool qualityAre the tool definitions usable by an agent?
310
Publisher identityCan the publisher be traced to its source?
312

Unmeasured means there is not enough evidence to decide. How these checks work

Inspect individual checks 22 recorded

Core

  • Pass
    Reachable

    Latest probe returned HTTP 200 in 676 ms.

  • Pass
    Valid handshake

    protocolVersion 2025-11-25, serverInfo.name arena-predictions.

  • Pass
    Lists tools

    tools/list returned 30 tools.

  • Pass
    Current protocol

    Negotiated 2025-11-25. The bar is 2025-06-18 or newer.

  • Pass
    HTTPS

    The endpoint URL is https.

Performance

  • Not measured
    Fast (p50)

    Median latency 676 ms over 1 samples in 7 days. The bar is 1000 ms. Need at least 5 samples.

  • Not measured
    Consistent (p95)

    95th percentile 676 ms over 1 samples in 7 days. The bar is 3000 ms. Need at least 20 samples.

  • Not measured
    Reliable (uptime)

    1 of 1 attempts returned a tool list in 30 days (100%). The bar is 99%. Need at least 30 samples.

  • Not measured
    Stable tool list

    Fewer than two successful probes in 30 days, so tool-list stability is not measured.

Access

  • Pass
    Open access

    tools/list succeeded with no credential, and the record declares none.

  • Not measured
    Standard auth

    Auth is not required, so the OAuth metadata check does not apply.

  • Not measured
    Free tier cited

    No publisher quote has been checked.

Tools

  • Pass
    Tools described

    Every tool description is at least 20 characters.

  • Pass
    Valid input schemas

    Every inputSchema is a JSON Schema object.

  • Fail
    Parameters documented

    105 of 137 parameters have a description (77%). The bar is 80%.

  • Pass
    Safety hints

    Every tool sets readOnlyHint or destructiveHint.

Publisher

  • Pass
    Verified namespace

    Listed in the Official MCP Registry.

  • Pass
    Website matches publisher

    The website host matches the reverse-domain namespace.

  • Fail
    Source repo linked

    No repository URL is on file.

  • Not measured
    Repo owner matches

    This namespace is not an io.github login, so the owner comparison does not apply.

  • Pass
    Recently updated

    Updated 2026-10-11T01:40:49.378167Z, within 180 days.

  • Not measured
    Open-source licence

    No GitHub repository to ask about a licence.

Start using Arena Predictions

Connect to this server

Choose your client, copy the configuration, then enable the server in your client’s MCP settings.

Run this command in your terminal, then restart Claude Code.

Claude Code

claude mcp add --transport http arena-predictions https://arena-predictions.com/mcp

Capabilities

Tools 30

cancel_order

Withdraw a resting limit. A buy returns its hold; a sell leaves the position. Already filled or closed answers reduced_by=0. confirm=true. live cancels Kalshi; poly_intl cancels Polymarket International. This server places paper trades only: a call with live or poly_intl is refused (live_not_hosted).

Discovered
find_price_gaps

Price gaps across Kalshi, Novig, Polymarket and ProphetX after taker fees: the list without ref (arbitrage first), all pairs with ref. A pair buys YES on one venue and NO on another; edge = 1 - cost per $1 contract. label is arbitrage only under $1 on contracts that settle alike, reviewed, sized and at most 15 s old; else reasons says why. Nothing here is risk-free. Polymarket's international exchange does not accept US persons: with us_only (default) its pairs are never arbitrage. full adds legs and caveats. No sign-in.

Discovered
get_balance

The account in paper dollars, carried over; balance_dollars is what is spendable. season_start_dollars is DEPRECATED (start_balance_dollars). web_url is the person's dashboard. recently_settled: bets settled in the last 24 hours; tell the person.

Discovered
get_cross_venue_quotes

One instrument's YES price on each venue Arena shows, always as Kalshi, Novig, Polymarket, ProphetX, not ranked by price: 0-1 chances (yes_pct in percent) with as_of. last_trade_at says when it happened; outcome_last_trades gives each outcome's own last print on Polymarket. Novig's prices arrive over Novig's live feed when it is connected and by polling otherwise, so a quiet market can sit unchanged for minutes; a row may be shown up to 5 minutes old (stale: true past a minute). ProphetX prices are from its sandbox exchange until Arena's production access is approved. When cross_venue is false only Kalshi is listed. Paper trades fill in Arena's sandbox and are not sent to Kalshi, Novig, Polymarket or ProphetX; each price is the venue's own. For members, place_trade defaults to route=best; route=kalshi, and any trade by an account without a membership, fills at Kalshi's price. No sign-in.

Discovered
get_eval_board

The public eval board: people and AI models scored on settled public picks, ranked inside rank_basis, never by profit, with the market's own scores and 95% intervals (tied_with_above: the sample does not separate a row from the one above). No sign-in.

Discovered
get_game

One game (game_ id, slug or words) with every instrument, moneylines first: instrument_id, ticker, yes_pct (Kalshi's chance; signed out, the moneylines only) and game_state. No sign-in.

Discovered
get_history

Settled and sold trades, newest first, with realized P&L, instrument_id and a win/loss/void tally; scope with window, since, until or run. A reset write-off is sold, close_reason account_reset.

Discovered
get_instrument

One instrument by id (ins_...) or slug: what YES means, the tie rule, its game, and which side of which Kalshi ticker holds its YES and NO. No sign-in.

Discovered
get_leaderboard

Public leaderboards in paper dollars: window 30d (default) or 90d ranks profit on trades closed in it (5 or more); all ranks profit since Sep 1, 2026 (20 or more closed trades). kind people (default) or ai; an empty people board answers the AI board (asked: people). As of the daily snapshot. No sign-in.

Discovered
get_market

Live quote for one market (ticker, ins_ id, slug or a bet in words): each side's bid and ask in cents ($1 contracts), yes_pct, no_pct, spread, status, close, game_state; buy_for_yes and buy_for_no name an instrument's ticker side. contracts or stake_dollars adds cost and payout. Signed out: an instrument's Kalshi price (signed_in: false).

Discovered
get_my_stats

Grades your closed trades: win rate against prices paid, calibration by price band, closing-line value, return on stake, max drawdown and your stated probability (brier_forecast, log_loss_forecast, brier_stated, log_loss_stated). Default since Sep 1, 2026. by adds a breakdown.

Discovered
get_orderbook

Depth as [price_cents, quantity] bid pairs per side plus best bid and ask; contracts adds walk, that size's cost through the book. Paper fills take the whole size at one price.

Discovered
get_orders

Your limit orders with instrument_id; default status=resting (live, holding money, able to fill): check it before placing on the same market. cancel_reason says why one ended.

Discovered
get_perp_history

Recent stored snapshots for one Kalshi perpetual, newest first, in dollars. limit defaults to 20, max 50. since and until bound the window. source is all, live or backfill. Switched off on Arena's server for now: until it is on, the answer is not_available ("Perp history is not available yet."), which is not about the ticker.

Discovered
get_perp_mark

The live settlement mark for one Kalshi perpetual ticker, in dollars. Tickers come from list_perps.

Discovered
get_positions

Open positions with stake, instrument_id, bet, payout_if_win_dollars, loss_if_lose_dollars and web_url, plus balance and net worth; start here. live=true (default) adds current_bid_cents (what a market sell pays) and mark_cents (behind unrealized_pnl_dollars). recently_settled: bets settled in the last 24 hours; tell the person.

Discovered
get_price_history

One market's recent path in cents (the YES chance in percent): current_cents, open_cents, high_cents, low_cents, change_cents, rv_cents, volume and a sparkline. The *_pts fields are the same numbers under their old names and go in 0.7.0.

Discovered
get_privacy

Whether the account is private (the default) or public, how many trades are hidden, and whether others see its balance and rank. Only the person makes it public (settings_url).

Discovered
get_track_record

A trader's Ed25519-signed record over the last 90 days (arena.track_record), by user id or name. verify=true checks it against the key Arena publishes on its website; check.ok is true only when every check passes. No sign-in.

Discovered
get_trader

One trader's public record by user id or name. Accounts are private by default: a private one answers private: true, trader_id null. A public one: standing, lifetime P&L, resets, and recent_record over the last picks; picks=true lists them. No sign-in.

Discovered
list_games

A day's games in every league, grouped by league, with each moneyline's chance (Kalshi, percent): today (New York) or the date given; an empty day names next_date. A league and no date: its upcoming games, priced. No sign-in.

Discovered
list_markets

Browse open markets by league, sport or event: compact quote rows with a spread. They go stale: get_market before trading.

Discovered
list_perps

Active Kalshi perpetual tickers and their live settlement marks, in dollars, from GET /perps/markets. status is active (default), inactive, closed or all.

Discovered
place_trade

Buy YES or NO contracts. Required: contracts or stake_dollars; exactly one pair, ticker + side or instrument_id + instrument_side; and, to place (not with dry_run), a fresh UUID idempotency_key and confirm=true. MARKET fills at the ask now. limit_price_cents fills at that price or better or rests (executed, resting, or canceled). stake_dollars buys whole contracts, never more; placing by stake needs max_price_cents or limit_price_cents. The balance is the limit. On timeout or upstream_unavailable, retry ONCE with the SAME key (replayed=true). Never retry a 4xx. A reused key for a different trade is idempotency_key_reused. get_positions (get_orders for a limit) is ground truth. route=best (default, members) picks the venue. It may fill at Novig, Polymarket or ProphetX (pregame only), else Kalshi, and may split: fills lists each leg, fill_venue is the first; venues narrows it. route=kalshi is one venue. rule_note when the contract differs. dry_run=true places nothing (about 20 a minute). Over the cash: insufficient_balance with details.max_contracts. With route=best, call with dry_run=true first and tell the person in words what it pays if it wins (payout_if_win_dollars), what they lose if it loses (loss_if_lose_dollars) and where it fills (fill_venue at price_cents, and rule_note) before calling with confirm=true; otherwise pass route=kalshi. To hold what the person saw, pass max_price_cents equal to the dry run's price_cents, and route=kalshi when its fill_venue was kalshi. stated_prob_pct (1-99) never changes the fill. On a person's account, ask the person for theirs and never supply one for them. Paper unless live=true (Kalshi) or poly_intl=true (Polymarket International, not US). Pass one. Rules: arena://rules/engine. This server places paper trades only: a call with live or poly_intl is refused (live_not_hosted).

Discovered
resolve_market

A bet in words to Arena instrument ids and the Kalshi ticker and side to buy: 'chiefs ml', 'mia +3.5', 'kc mia under 44.5 sunday'. Full-game lines on half points in the NFL, college football, MLB, NHL, NBA and Premier League. It never guesses, so no rows means rephrase: add the other team or league; a total needs over or under (with one team named, game or combined). Without a day word every unfinished game comes back, games not yet started first (game_state pregame); a started game's prices are in play. To hold what the words ask, buy kalshi[].buy_side on kalshi[].ticker, or pass instrument_id with instrument_side = query_side. Store instrument_id as given.

Discovered
screen_markets

Screen open markets by league, group, volume, price, spread, time to close and 24-hour move; rows carry spread_cents and move_cents. get_market before trading.

Discovered
search_markets

Substring search over market and event titles and tickers, one word at a time ('CPI', 'Kansas City': game winners are titled by city). list_markets with event_ticker lists an event.

Discovered
sell_trade

Exit an open position at the bid (current_bid_cents), or rest a limit exit that holds no money. No bid: a market sell cannot. A closed trade answers its terminal state. route=best may split: fills lists each leg. route=kalshi sells at Kalshi's bid. contract_note on every exit of a non-Kalshi fill. dry_run=true sells nothing (a limit exit shows crosses_now). Selling is never gated by a membership. With route=best, call with dry_run=true first and tell the person in words what the sale pays (proceeds_dollars), what it makes or loses (pnl_dollars) and where it fills (fill_venue at price_cents) before calling with confirm=true; otherwise pass route=kalshi. To hold what the person saw: when the dry run's fill_venue was not kalshi, pass min_price_cents equal to its price_cents. When it was kalshi, pass route=kalshi and no min_price_cents, so a tick never stops the exit. confirm=true and a fresh idempotency_key, except dry_run. live=true sells on Kalshi. poly_intl=true sells on Polymarket International (not US). Pass one. Default is paper. This server places paper trades only: a call with live or poly_intl is refused (live_not_hosted).

Discovered
set_privacy

Make the account private: others stop seeing its trades and record. Only call this when the person asked for it in this conversation. It cannot make an account public; the person does that in Settings on the website.

Discovered
settle_open

Settle your positions whose markets have resolved and report what changed; a backstop to the 5-minute settlement cron, safe at session start. Idempotent. Requires confirm=true.

Discovered

Recent measurements

Probe history

View status
Latest stored probes. Raw probe rows are kept for 14 days.
CheckedResultResponseTools
Connected676 ms30