MCP.Market

PredMCP MCP server

online · 100% uptime
by RavioleLabs·io.github.RavioleLabs/predmcp·v0.5.0
No reviews yet

Polymarket + Hyperliquid + macro for AI agents. 38 tools, signal backtest, SSE streaming. Free tier.

B76/100grade B
Write a reviewRun it
What users say
No reviews yet
Be the first
Safety scan
B76/100

full report

Adoption
New

0 stars

Reviews

Write one

Nobody has reviewed PredMCP yet.

If you have run it, two minutes of your experience saves the next person an afternoon.

PredMCP tools (52, 1 write)

write = sends, deletes, buys or posts
  • create_api_keywrite actionFree

    Generate a free PredMCP API key. Requires an email. Returns the key and ready-to-use MCP config. Free tier: 100 calls/day, one key per IP.

  • get_basic_macroFree

    DXY, US10Y yield, S&P 500, gold, VIX — direct from free Yahoo Finance. Raw values + 1d change. Pro adds BTC dominance, ETH/BTC, total crypto mcap, and a RISK_ON/OFF regime classifier.

  • get_carry_scannerFree

    Funding carry NET of costs: annualized funding minus (spread + 2x slippage at your size), with break-even holding period and a 7d funding-stability score from our snapshots. get_top_funding_rates shows gross — nobody trades gross.

  • get_cex_outflowsFree

    Net ETH outflows across known CEX hot wallets (Binance, Coinbase, OKX, Kraken, Bitfinex) over a window. Outflow = bullish (BTC moving to cold storage). Inflow = distribution pressure. Etherscan free tier.

  • get_conviction_scoreFree

    Aggregates funding outlier, whale imbalance, OI/volume ratio, and momentum into a single directional score (-100 short ↔ +100 long) plus a 0-100 strength score. One call replaces 6+ lookups, returns a clean number for LLM decisions.

  • get_cross_venue_fundingFree

    Predicted funding spreads between Hyperliquid, Binance, and Bybit per asset (from HL's predictedFundings feed). Surfaces delta-neutral carry: long the venue with lowest funding, short the highest, collect the spread. Sorted by annualized spread.

  • get_funding_curve_anomalyFree

    Term-structure analysis of Hyperliquid funding for an asset: current vs 8h avg vs 24h avg vs 7d baseline. Flags spikes, regime shifts, sign contradictions. More nuanced than raw funding rate.

  • get_funding_outliersFree

    Hyperliquid perps whose current funding rate deviates significantly from their 7-day average. A spike vs baseline is a stronger signal than raw rate.

  • get_funding_ratesFree

    Current funding rates for Hyperliquid perpetuals. Positive rate = longs pay shorts (bearish bias); negative = shorts pay longs (bullish bias).

  • get_hip4_vs_pm_arbFree

    Finds the same underlying market priced on both HIP-4 (on-chain Hyperliquid) and Polymarket, flagging spreads above threshold. A spread means one venue is mispriced relative to the other.

  • get_late_game_sportsFree

    Sports prediction markets on Polymarket closing within a few hours with a high-certainty leading outcome. Targets near-certain resolution for late-game positioning.

  • get_liquidation_clustersFree

    Estimated price levels where mass liquidations concentrate for a given Hyperliquid perp, computed from mark price and standard leverage multiples. Higher nearby orderbook liquidity = stronger support/resistance.

  • get_macro_contextFree

    Live macro snapshot: DXY, US10Y yield, S&P 500, gold, VIX (Yahoo Finance free) + BTC dominance, ETH/BTC, total crypto market cap (CoinGecko free). Plus a coarse RISK_ON / RISK_OFF / MIXED regime.

  • get_macro_liquidityFree

    Fiat-to-crypto liquidity gauge: BTC + ETH spot ETF flows (Farside) plus on-chain USDT + USDC mint/burn (Etherscan). Net inflow = bullish for risk assets.

  • get_market_contextFree

    Unified intelligence snapshot for any topic, asset, or keyword: all matching Polymarket and HIP-4 prediction markets combined with live Hyperliquid perp data (price, funding, OI). One call replaces 3+ separate lookups.

  • get_market_edgeFree

    Universal fair-value anchor for ONE market of ANY category. Pass the question (+ optional strike/comparator/expiry/underlying for price markets, + yes_price). Dispatches to the best anchor: options/vol fair value (crypto price), a confident Polymarket twin (events/politics/sports), or a soft news lean. Returns { method, confidence, anchor_prob, hip4_yes, edge_pt, verdict, source_ref, note }. edge_pt = (anchor_prob - hip4_yes)*100; positive = YES cheap; null for untraded/synthetic prices.

  • get_market_edgesFree

    One-call universal mispricing scanner across ALL HIP-4 market types (crypto price, sports, politics, macro, events). Resolves every market to its best anchor (options / Polymarket twin / news lean / none) and ranks by |edge|. Filter with method / min_edge_pt / limit. Returns by_method counts + ranked markets. Superset of get_outcome_edges.

  • get_market_regimeFree

    One-call market regime classifier: RISK_ON_TRENDING / RISK_OFF / SQUEEZE_RISK / CHOP_LOW_VOL / MIXED. Combines BTC trend + realized vol, Deribit IV premium, funding breadth, and OI-weighted crowding across the top 50 perps. Call this FIRST each session to condition strategy choice.

  • get_marketsFree

    Live prediction markets from Polymarket and/or HIP-4, sorted by volume. Returns title, YES/NO prices, 24h volume, and expiry.

  • get_markets_near_resolutionFree

    Polymarket markets resolving within the next N hours with a leading probability above threshold. Useful for resolution arbitrage and last-minute positioning.

  • get_moversFree

    Top prediction markets ranked by 24h volume spike or biggest YES/NO price swing. Surfaces breaking news bets and momentum plays across Polymarket and HIP-4.

  • get_news_correlationFree

    Recent crypto news headlines mentioning an asset (CoinDesk, The Block, Decrypt, Cointelegraph) paired with the 1h price move that followed each. Lets agents filter news bias.

  • get_news_feedFree

    Recent headlines across crypto, finance and sports RSS feeds, unfiltered and unclassified. Unlike get_recent_news it does not filter by asset ticker, so it also covers the sports, macro and event markets that make up most of a HIP-4 board. Match the headlines against your own market list.

  • get_news_signalsFree

    News → markets recommender: pulls recent crypto headlines, classifies each event type (depeg/hack_exploit/listing/delisting/unlock/regulatory/partnership/outage/hype/fud/macro), extracts the assets, maps to affected HIP-4 markets, and assigns a directional lean (YES up / YES down). Heuristic, low/medium confidence, NOT financial advice; war/hard-politics excluded. Powers a live "event → affected markets → lean" feed. Filter with hours / event_type / min_confidence / limit.

  • get_oddsFree

    Current YES/NO prices and implied probability for any Polymarket or HIP-4 market token.

  • get_oi_divergenceFree

    Price-vs-OI regime per coin from our continuous snapshots: NEW_LONGS / SHORT_SQUEEZE / NEW_SHORTS / LONG_LIQUIDATION. Scan all tracked coins or analyze one. Impossible without OI history, which no public API provides — only predmcp records it.

  • get_oi_historyFree

    Open-interest time series for a coin over the last 24h, from our continuous 5-min collector. Hyperliquid has NO OI-history endpoint — this data exists only on predmcp. Includes price + funding at each point. Pro tool get_oi_divergence classifies price-vs-OI regimes (squeeze/liquidation/new positioning) across all coins.

  • get_oi_near_capFree

    Lists Hyperliquid perps that are currently at the open interest cap — new long positions cannot be opened. Use as a blacklist to avoid getting rejected on entry.

  • get_options_ivFree

    BTC or ETH options snapshot via free Deribit feed: ATM implied volatility, put-call skew, term structure, total OI and 24h volume. Useful as a sentiment gauge — IV high = market expects big moves; skew indicates direction of hedging.

  • get_orderbookFree

    Full orderbook depth (bids + asks) for any Polymarket market token. Shows liquidity at each price level.

  • get_orderbook_depthFree

    Full orderbook depth + slippage estimate for any Hyperliquid perp or HIP-4 market. Returns top of book, spread, cumulative depth at $100/$500/$1k/$5k tiers, and estimated slippage for a given order size. Critical for HIP-4 farming and low-liquidity assets where market orders get destroyed by slippage.

  • get_outcome_edgesFree

    One-call mispricing scanner for autonomous agents: every live HIP-4 PRICE market, each priced against its options/vol-implied fair value and ranked by |edge|. BTC/ETH use the Deribit IV smile; other assets use Hyperliquid realized vol (iv_source labels which). edge_pt = (implied_prob - yes_price)*100, computed only where a real traded YES price exists (synthetic/untraded 0.5 placeholders return edge_pt:null so no fake edge is surfaced). Filter with min_edge_pt / underlying. The whole board without looping.

  • get_outcome_fair_valueFree

    Options/vol-implied fair value for a HIP-4 binary/range/touch PRICE market ("<asset> above $X at time T?"). Returns the risk-neutral probability the market resolves YES, and — if you pass market_yes_price — the signed edge in probability points and a YES cheap/rich/fair verdict. BTC/ETH price off the live Deribit IV smile (interpolated to strike + DTE); ANY other asset falls back to Hyperliquid 30d realized vol. iv_source labels which was used (smile/term_interp/skew_adj/atm/realized_vol). The touch case uses the correct drifted first-passage formula. Approximations are listed in caveats.

  • get_pm_hl_divergencesFree

    Markets where Polymarket implied probability diverges from Hyperliquid perpetual funding direction — e.g. PM prices bullish outcome but HL funding shows crowded longs (bearish pressure). The hardest signal to compute manually.

  • get_portfolio_riskFree

    Risk analytics for a list of positions: per-position beta to BTC and ETH, sector concentration, pairwise correlation matrix, portfolio annualized volatility, 1-day 95% parametric VaR. Uses 14d of 1h Hyperliquid candles.

  • get_position_sizeFree

    Turn a signal + bankroll into a concrete position: fractional-Kelly size capped by orderbook liquidity, ATR-based stop, liquidation price at chosen leverage, funding carry cost, and % bankroll at risk. Warns when liquidation sits inside the stop. Ground win_rate_pct with get_signal_performance or get_signal_backtest first.

  • get_price_summaryFree

    One-call snapshot for an asset: mark price, 24h and 7d returns, 30d high/low, drawdown from high, rally from low, annualized realised volatility. Computed from 30d of HL hourly candles.

  • get_recent_newsFree

    Recent crypto news headlines matching an asset ticker, from CoinDesk + The Block + Decrypt + Cointelegraph RSS feeds. Pro adds the 1h price move that followed each headline.

  • get_recent_signalsFree

    Server-detected events from the last hour: funding outliers (≥3x 7d baseline), whale trades (≥$100k), OI caps reached. Cursor-based — pass next_cursor back as since_id to receive only new events. The polling equivalent of the /sse/signals stream. Pro tool get_signal_history covers 7 days with forward-return outcomes.

  • get_setup_qualityFree

    Execution-quality score (0-100, A-F grade) for entering a position right now: spread, slippage at desired size, orderbook depth, volatility regime, trend alignment, support/resistance proximity. Different from conviction (direction) — this scores ENTRY conditions.

  • get_signal_backtestFree

    Find historical instances of a signal type on an asset over the last N days and compute forward returns (1h/4h/24h), win rate, and Sharpe. Lets an agent reason about EV before trading. Killer feature: turns predmcp from data API into edge-proven intelligence.

  • get_signal_historyFree

    Server-detected signal events over up to 7 days (funding outliers, whale trades ≥$100k, OI caps), each joined with its measured forward returns (1h/4h/24h) once mature. Cursor-based. "What happened last time funding spiked on HYPE — and did it matter?" in one call.

  • get_signal_performanceFree

    Hit rates and average forward returns per signal type, measured on OUR actually-emitted production signals (not a backtest reconstruction). Use this to ground win_rate inputs for get_position_size. Sample sizes included — treat n < 10 as anecdotal.

  • get_simple_ivFree

    BTC or ETH options snapshot via free Deribit public feed: ATM implied volatility, total OI, 24h volume. Pro adds put-call skew + full term structure parsing.

  • get_top_funding_ratesFree

    Top Hyperliquid perps ranked by absolute funding rate, with OI and annualized yield. Useful for finding the most overcrowded longs/shorts and carry opportunities.

  • get_upcoming_catalystsFree

    Lists upcoming market catalysts for an asset within a horizon: token unlocks, governance votes (Tally), ETF/SEC deadlines, FOMC dates. Helps agents avoid blind trades into known events. Free data sources (Tally + curated DB).

  • get_volume_spikesFree

    Polymarket markets with abnormal 24h volume vs their 7-day daily average. Volume spikes typically precede news events or informed positioning.

  • get_whale_flowFree

    Cumulative whale buy/sell imbalance over hours-to-days from the durable trade tape (≥$25k trades persisted continuously, restart-proof). Returns imbalance ratio (-1..+1), BUY/SELL_DOMINANT verdict, and the largest recent prints. Longer horizons than get_whale_trades (1h buffer).

  • get_whale_labelFree

    Look up an Ethereum address against our curated label DB (CEX hot wallets, known market makers, suspected funds). Lets agents distinguish mechanical MM flow from alpha-generating activity.

  • get_whale_positionsFree

    Positions of a specific Polymarket wallet, optionally filtered to one market. The /positions API requires a user (no public "all holders by market" endpoint), so pass the wallet address of the trader you want to inspect.

  • get_whale_tradesFree

    Recent large trades on Hyperliquid perps above a notional threshold. Includes side (long/short), size, price, and timestamp.

  • search_marketsFree

    Full-text search across all Polymarket and HIP-4 prediction markets. Returns ranked results with current odds.

Public scan report

scanner v0.1.9 · 2026-09-27 · same rubric, same numbers if you re-run it

no findings
  • –Code scanremote-only server, no package to scann/a
  • Live reliabilityremote reachable in 1492ms20/20
  • Tool poisoning52 tool descriptions checked15/15
  • Auth qualityAPI key sent as a header8/15
  • Maintenancelast push 109 days ago8/15
  • Maintainer identityregistry namespace matches repository owner6/10
Overall 76/100. Components that don't apply are left out of the denominator. Any critical finding is an F.RubricAppeal a findingJSON

How PredMCP compares

Against every graded MCP server, from our own scans and uptime checks. Updated after each scan run.

Safety score
76/100
higher than 48% of 48,255 servers
Tools
52
the typical server has 7
Tools that change things
2%
typical: 0%. Fewer write tools means less an agent can break
Uptime
100%
of our recent checks; the typical remote server answers 100%
Response time
75 ms
faster than 89% of 30,031 servers

What its tools let an agent do: 51 read · 1 create

Install PredMCP in Claude Code, Cursor or VS Code

claude mcp add --transport http predmcp https://predmcp.com/mcp
Add to Cursor

What the publisher says

From the PredMCP repository's README, as published. We do not edit it. Read it on GitHub

PredMCP

Safe, read-only market data for AI trading agents.

44 MCP tools that cross Polymarket prediction markets, Hyperliquid perps, and HIP-4 native predictions. Plug your agent into live trading data — without ever giving it the ability to execute orders.

Live, hosted, free during early access: predmcp.com

Why read-only matters

The #1 worry when wiring an LLM to a trading venue is the same in every conversation: "and it can't actually trade, right?" PredMCP is built so that question has a single answer.

  • No execution endpoint. Zero order-placement code exists in this codebase. Not gated, not commented out — not implemented. There is nothing to enable.
  • No private keys touched. We use the public Hyperliquid Info API and Polymarket Gamma API. Your wallet, your seed phrase, your accounts — never seen, never asked for.
  • 44 tools, all queries. Markets, orderbooks, funding rates, whale activity, signals. Every tool is a read. Your agent reasons about markets; you stay the only one who can act on them.

What's in the box

Public data (any free key)

Intelligence (early access — free during the open phase)

Each tool is one verb your agent can call. Self-documenting via MCP.

Examples

Iran — what prediction markets say right now:

"US x Iran peace deal by May 15"    → YES 20%  ($9M volume)
"US x Iran peace deal by May 31"    → YES 34%  ($16M volume, +4% today)
"Iranian regime falls before 2027"  → YES 15%  ($17M volume)
"US invades Iran before 2027"       → YES 21%  ($26M volume)

One getmarketcontext call.

BTC cross-platform signal:

HL perps:  $79,740  |  funding: neutral  |  OI: 31,100 BTC
HIP-4:     BTC > $81,041 by 6am tomorrow  →  YES 12%  (market says no)
Signal:    ✓ ALIGNED — perps neutral, prediction bearish

When these diverge, it's a signal. get_signals computes it.

24h movers:

"Will Bitcoin hit $150k by June 30?"   $5.8M traded in 24h
"US x Iran permanent peace deal"       $3.2M in 24h  (+4% move)
Athletics vs Phillies                  +55% price move

Use the hosted server (recommended)

Get a free API key (email, no credit card, 100 calls/day) at predmcp.com/signup, then drop into your claudedesktopconfig.json:

{
  "mcpServers": {
    "predmcp": {
      "type": "http",
      "url": "https://predmcp.com/mcp",
      "headers": { "x-api-key": "YOUR_KEY" }
    }
  }
}

Also works with Cursor, Windsurf, Claude.ai (OAuth), and any MCP client that supports HTTP transport.

Ask your agent:

  • "What are prediction markets saying about Iran right now?"
  • "Is there a divergence between BTC perp traders and prediction markets?"
  • "What markets moved the most in the last 24 hours?"
  • "Are any whales active on both Hyperliquid and Polymarket for BTC right now?"

Shortened. The full README is on GitHub.

Nothing above is checked by us. What we check is on the safety report.

PredMCP: common questions

Is PredMCP MCP server safe?
Mostly: it is graded B (76/100). Read the PredMCP safety report
How do I install PredMCP?
It runs remotely at predmcp.com. Add it to Claude Code, Cursor, VS Code or Claude Desktop with the snippets above, or call it through the mcp.market gateway without installing anything.
Does PredMCP need an API key?
Yes. The registry entry asks for x-api-key.
Is PredMCP maintained?
The last commit was 147 days ago (2026-06-10). The latest release is v0.5.0.
Is PredMCP up?
100% of our last 28 checks got an answer. We check remote servers about four times a day.
How does PredMCP compare with other MCP servers?
safety score 76/100 (higher than 48% of 48,255 servers); tools 52 (the typical server has 7); tools that change things 2% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.