Prediction market intelligence, inside your AI agent.

Consensux starts from markets where real money is on the line — then cross-checks them against options flow, macro signals, public disclosures and forecasting platforms to detect consensus shifts, mispricings and divergences before they become obvious.

Delivered as callable MCP tools inside Claude, Cursor, Continue and any MCP-compatible client.

claude · consensux-mcp

❯ Will the Fed cut rates in September?
→ consensux.get_context_summary({ event: "Fed rate cut, Sep 2026" })
{
  "event_id":      "fed-cut-sep-2026",
  "consensus":     0.57,
  "fair_value":    0.61,
  "verdict":       "underpriced",
  "lean":          "yes",
  "edge_pts":      4,
  "confidence":    "medium",
  "sources":       { "polymarket": 0.61, "kalshi": 0.52, "manifold": 0.66, "metaculus": 0.55 },
  "cross_signals": { "cme_rates": 0.60, "cboe_options": 0.61, "fed_minutes_nlp": "dovish" }
}

get_consensus_score get_divergence_alerts get_trend_velocity get_history compare_signals get_context_summary search_events

Trusted market-implied odds

Prediction markets are one powerful signal — not gospel. Consensux weights Polymarket, Kalshi, Manifold and Metaculus into a single market-implied probability you can rely on.

Early divergence detection

Compare prediction markets with other signals — and get alerted when they start to diverge.

Agent-ready, less research

Structured, sourced answers your agent calls directly via MCP — minutes of tab-juggling collapsed into one tool call, in Claude, Cursor, Continue or any MCP client.

Live track record

Signal accuracy

Real, sourced, never back-filled. Calibration accrues as the markets we track settle.

—

Calls correct

—

▲ Underpriced

—

▼ Overpriced

—events tracked —data points —resolved

A needle past the 50% mark = the tool's calls were right more often than a coin flip. Correct means the market resolved the way the verdict implied (underpriced → YES, overpriced → NO).

Private beta · accruing

From market odds to actionable signals.

When people risk their own capital, they verify every assumption and correct their own biases — or lose money. Pundits face no such pressure. It's the same edge the Good Judgment Project documented under IARPA's ACE program: accountable, incentivised crowds consistently outforecast expert analysts.