Skip to main content
# Install in one command
uvx rekko-mcp

​
What this page covers

  • What MCP is and why it matters for prediction market research
  • Installing and configuring rekko-mcp for Claude Code and Cursor
  • Workflow examples: screening, analysis, arbitrage, portfolio
  • Available MCP tools (25+)
  • When to use MCP vs the REST API

​
What is MCP?

The Model Context Protocol (MCP) lets AI coding assistants call external tools during a conversation. Instead of copy-pasting API calls, you describe what you want in natural language and the assistant uses the right API endpoints automatically. Rekko’s MCP server (rekko-mcp) exposes 25+ tools covering market data, deep analysis, trading signals, arbitrage detection, and portfolio management. Your AI assistant calls these tools behind the scenes while you focus on the research question.

​
Installation

​
Requirements

​
Claude Code

claude plugin install rekko

​
Cursor

Create or edit .cursor/mcp.json in your project root: 
{
  "mcpServers": {
    "rekko": {
      "command": "uvx",
      "args": ["rekko-mcp"],
      "env": {
        "REKKO_API_KEY": "rk_free_your_key_here"
      }
    }
  }
}

​
Verify installation

Ask your assistant: “List the top 5 prediction markets by volume.” If rekko-mcp is configured correctly, it will call the list_markets tool and return formatted results.

​
Workflow examples

​
1. Market screening

You say: “Find Kalshi markets with high volume that might have mispriced odds.” What happens behind the scenes:
  1. Assistant calls list_markets(source="kalshi", limit=20) to fetch active markets
  2. Calls screen_markets(...) to score them by volume, movement, and category
  3. Presents the top candidates with scores and recommended actions
Example output: 
Top Kalshi markets by opportunity score:

1. KXFED-26APR30 — "Fed rate decision April 2026"
   Score: 82 | Volume: $1.2M | Action: analyze

2. KXBITCOIN-26APR — "Bitcoin above $150K by April?"
   Score: 74 | Volume: $890K | Action: analyze

3. KXINFLATION-26MAY — "CPI above 2.5% in May?"
   Score: 68 | Volume: $540K | Action: watch

​
2. Deep analysis

You say: “Analyze the Fed rate decision market on Kalshi. I want to understand the key factors driving the price.” What happens:
  1. Calls analyze_market(bet_text="Fed rate decision April 2026", platform="kalshi")
  2. Returns an analysis_id — the research pipeline takes 30-90 seconds
  3. Polls check_analysis_status(analysis_id) until complete
  4. Retrieves the full analysis with probability, confidence, edge, key factors, and risks
Example output: 
Analysis complete for KXFED-26APR30:

Probability: 71% (market: 62%) — Edge: +9 points
Confidence: 82% | Risk: low
Recommendation: BUY YES

Key factors:
- PCE inflation at 2.1% — within Fed's comfort zone
- Three FOMC members signaled openness to cuts
- Jobless claims trending upward (225K → 238K)

Risks:
- Strong employment report could reverse dovish signals
- Tariff escalation could reignite inflation expectations

Sources analyzed: 14

​
3. Arbitrage hunting

You say: “Are there any arbitrage opportunities between Kalshi and Polymarket right now?” What happens:
  1. Calls get_arbitrage(min_spread=0.02) for cached results (instant)
  2. If you ask for fresh data, calls get_arbitrage_live(min_spread=0.02) (10-30 seconds)
  3. Returns matched events with spreads, scores, and which platform is cheaper

​
4. Portfolio management

You say: “Show my shadow portfolio and suggest what to do next.” What happens:
  1. Calls get_portfolio(mode="shadow") to list open positions
  2. Calls get_performance(mode="shadow") for aggregate stats
  3. May call check_resolutions() to settle any resolved markets
  4. Presents positions, P&L, win rate, and suggestions

​
5. Trading signals

You say: “Should I buy the Bitcoin 150Kmarket?Ihavea150K market? I have a 10K bankroll and already hold positions in the Fed rate market.” What happens:
  1. Calls get_portfolio_strategy(...) with your current positions and bankroll
  2. Returns a portfolio-aware signal accounting for correlation between positions
  3. Includes Kelly-optimal size, hedge recommendations, and concentration warnings

​
6. Market data refresh

You say: “Refresh the Kalshi market data.” What happens:
  1. Calls run_scraper(platform="kalshi") to pull the latest markets
  2. Fresh data is available for subsequent queries

​
Available tools

​
Market intelligence

ToolDescription
list_marketsBrowse active markets, filter by platform
get_marketGet a single market with pricing data
search_marketsSearch markets by text query
get_market_historyPrice and volume history (48h, 7d, 30d)
get_resolutionResolution timing and mechanism
get_execution_guidanceOptimal trade execution analysis

​
Screening and discovery

ToolDescription
screen_marketsBatch score markets with volume and movement filters
get_calibrationSignal accuracy metrics (free, no auth)

​
Deep research

ToolDescription
analyze_marketTrigger a deep AI analysis pipeline (async)
check_analysis_statusPoll analysis progress
get_analysisRetrieve completed analysis
list_analysesBrowse recent analyses

​
Strategy and portfolio

ToolDescription
get_strategyTrading signal with Kelly sizing (blocks 30-90s)
get_portfolio_strategyPortfolio-aware signal with correlation
get_consensusAggregated agent consensus probability

​
Arbitrage

ToolDescription
get_arbitrageCached cross-platform opportunities
get_arbitrage_liveFresh scan (10-30 seconds)

​
Correlation

ToolDescription
get_correlationCross-market correlation analysis

​
Trading

ToolDescription
place_shadow_tradePaper trade for tracking
report_tradeReport a trade for consensus aggregation
get_portfolioCurrent positions and P&L
get_performanceWin rate, ROI, and edge metrics
check_resolutionsSettle resolved markets

​
Data refresh

ToolDescription
run_scraperRefresh market data from a platform

​
Webhooks

ToolDescription
create_webhookRegister event webhook
list_webhooksList your webhooks
delete_webhookRemove a webhook

​
MCP vs REST API

ScenarioUse MCPUse REST API
Interactive research in your IDEYes
Automated trading botYes
Quick market check during codingYes
Production pipelineYes
Portfolio reviewYes
Webhook/streaming integrationYes
One-off analysis during developmentYes
Scheduled batch processingYes
MCP is ideal for exploratory research and ad-hoc queries. The REST API is better for automated systems that run without human interaction.

​
Tips for effective MCP usage

Be specific about platforms. “Show me Kalshi markets” is faster than “show me markets” because the tool skips Polymarket data. Request expansions explicitly. “Analyze this market and show me the causal factors” triggers the ?expand=causal parameter automatically. Chain analyses. “Analyze the top 3 markets from that screening” works — the assistant remembers prior results and makes sequential tool calls. Use shadow trading. “Place a shadow trade on the Fed market, $500 on YES” lets you track performance without risking real money.

​
What’s next

MCP tools reference

Full parameter documentation for all 25+ tools.

Build a trading bot

When you are ready to automate, use the REST API.

REST API quickstart

Get an API key and make your first call in 5 minutes.