About MCP Server Kalshi
An MCP server for Kalshi prediction markets that lets Claude Code and other agent harnesses browse markets, research data, read settlement rules and place trades. It suits users who need market access and trading through MCP tools. It targets the demo sandbox by default, and order-placing tools return a preview unless confirm is set to true.
Key features
- 01Query tools for markets, events and series
- 02Orderbook, candlestick and trade history research
- 03Settlement rules with contract PDF text extraction
- 04Exchange status and trading schedule lookup
- 05Portfolio balance, positions and fills
- 06Order placement, cancellation, amendment and reduction tools
Use cases
- 01Researching order books and price history before trading
- 02Reading settlement rules to confirm resolution conditions
- 03Reviewing account balance, positions and fills
- 04Testing trading workflows in the demo environment
MCP support
This is an MCP server that exposes Kalshi market, rules, portfolio and order tools to MCP clients.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx mcp-server-kalshiKALSHI_ENVdemo (sandbox) or prod (real money)KALSHI_API_KEYSecretKalshi API key ID; required only for portfolio/order toolsKALSHI_PRIVATE_KEY_PATHSecretPath to your RSA private key .pem; required for authenticated toolsBASE_URLOptional explicit REST base override (must include /trade-api/v2)Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEMCP Server Kalshi
An MCP server that gives Claude Code and other agent harnesses a first-class interface to Kalshi. It is built for end-to-end trading: browse markets, research them, read the exact settlement rules (including pulling the contract-terms PDFs), and execute trades — all through MCP tools.
Highlights
- Discovery —
list_markets,get_market,list_events,get_event,list_series,get_series. (Kalshi has no free-text search;list_marketsfilters are the search.) - Research —
get_market_orderbook,get_market_candlesticks,get_market_trades. - Deep rules —
get_market_rulesconsolidates a market'srules_primary/rules_secondary, early-close conditions, settlement sources, and series prohibitions;fetch_rules_pdfdownloads and extracts the text of the actual legal contract PDF so the agent can read it. - Exchange —
get_exchange_status,get_exchange_schedule(is the market open, and its hours). - Portfolio —
get_balance,get_positions,get_fills,get_settlements. - Trading —
create_order,cancel_order,amend_order,decrease_order, pluslist_orders/get_order.
Safety by default
- The server targets Kalshi's demo (sandbox) environment unless you explicitly set
KALSHI_ENV=prod. - Order-placing tools (
create_order,amend_order) requireconfirm=true. Without it they return a preview — a human-readable summary and the exact payload — and place nothing. - Credentials are optional: all market/rules tools work unauthenticated. Only portfolio and order tools need an API key + RSA private key.
Intuitive order model
Kalshi's V2 order API quotes everything from the YES leg (bid/ask in fixed-point dollars).
This server exposes the natural model instead — action (buy/sell) + side (yes/no) + a whole
cents limit price — and translates it (including the buy-NO ⇄ sell-YES price inversion).
Configuration
| Variable | Default | Purpose |
|---|---|---|
KALSHI_ENV | demo | demo (sandbox) or prod (real money). Derives the base URL. |
KALSHI_API_KEY | (none) | Kalshi API key ID. Required only for authenticated tools. |
KALSHI_PRIVATE_KEY_PATH | (none) | Path to your RSA private key .pem. Required for authenticated tools. |
BASE_URL | (derived) | Optional explicit REST base override (must include /trade-api/v2). |
See .env-example. Get API credentials at
docs.kalshi.com/getting_started/api_keys
and a demo account via the
demo environment guide.
Claude Desktop (uvx)
"mcpServers": {
"kalshi": {
"command": "uvx",
"args": ["mcp-server-kalshi"],
"env": {
"KALSHI_ENV": "demo",
"KALSHI_API_KEY": "<YOUR KALSHI API KEY>",
"KALSHI_PRIVATE_KEY_PATH": "PATH TO YOUR RSA KEY FILE"
}
}
}
Claude Desktop (Docker)
"mcpServers": {
"kalshi": {
"command": "docker",
"args": ["run", "--rm", "-i",
"--mount", "type=bind,src=/Users/username,dst=/Users/username",
"-e", "KALSHI_ENV", "-e", "KALSHI_API_KEY", "-e", "KALSHI_PRIVATE_KEY_PATH",
"mcp-server-kalshi"],
"env": {
"KALSHI_ENV": "demo",
"KALSHI_API_KEY": "<YOUR KALSHI API KEY>",
"KALSHI_PRIVATE_KEY_PATH": "PATH TO YOUR RSA KEY FILE"
}
}
}
Local Development
- Create a
.envfile (see.env-example). - Install deps:
uv sync(add--extra devfor dev tools). Requires Python 3.10+. - Run:
uv run start. - Test:
uv run pytest.
MCP Inspector
npx @modelcontextprotocol/inspector uv --directory /path/to/mcp-server-kalshi run start
Testing & code quality
uv sync --extra dev # install dev tools (ruff, mypy, pytest, ...)
uv run pytest # run the test suite
uv run pytest --cov # tests with a coverage report
uv run ruff check src tests # lint
uv run black src tests # format (add --check to verify only)
uv run mypy # type check
uv run pre-commit install # (once) run ruff + black on every commit
Tests are pure/offline — they exercise the order translation and confirm-gate, the HTTP client
(via an injected httpx.MockTransport), the MCP tool registry, config, and PDF extraction, all
without touching the live Kalshi API. CI (.github/workflows/ci.yml) runs ruff + black + mypy +
pytest across Python 3.10–3.13 on every push/PR, and releases are gated on that same suite.
Authentication
Requests are signed with RSA-PSS (MGF1-SHA256, max salt). Each authenticated request sends
KALSHI-ACCESS-KEY, KALSHI-ACCESS-TIMESTAMP, and KALSHI-ACCESS-SIGNATURE, where the signed
message is timestamp_ms + METHOD + path (path includes /trade-api/v2, excludes the query
string).
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.