Kalshi Kernel

Kalshi Kernel

An unofficial, safety-focused Model Context Protocol (MCP) server for Kalshievent contracts and perpetual markets. The server is designed for agentharnesses such as Hermes and exposes a deliberately small tool surface generatedfrom Kalshi's published API contracts.

Canonical publisher: CopyleftDev. Plannedsource and container releases use github.com/copyleftdev/kalshi-kernel andghcr.io/copyleftdev/kalshi-kernel respectively.

[!CAUTION]Pre-release software—not ready for trading. Version 0.1.1 is aconformance-tested scaffold. Only kernel_status is operational. Market-data,portfolio, paper-trading, and live-trading adapters are not connected and failclosed with capability_not_ready.

[!IMPORTANT]This community project is not an official Kalshi product and is not affiliatedwith, endorsed by, sponsored by, or supported by KalshiEX LLC or itsaffiliates. “Kalshi” is used only to identify interoperability with Kalshi'spublished APIs.

Why this project exists

Trading APIs are broad and optimized for application developers. Agents need anarrower boundary with explicit intent, strict schemas, predictable errors, andstrong separation between simulation and real-money execution. Kalshi Kernelprovides that boundary while continuously testing its generated clients againstthe latest upstream contracts.

The design priorities are:

• paper mode by default;
• immutable execution mode for the lifetime of a process;
• explicit acknowledgement and credentials before live mode can arm;
• fixed-point strings for prices and quantities;
• strict, generated tool schemas with safety annotations;
• fail-closed behavior when a capability is unavailable;
• traceable alignment with exact OpenAPI operations and AsyncAPI channels; and
• repeatable MCP protocol, race, fuzz, and upstream-drift testing.

Current status

Do not advertise, deploy, or rely on this version as a functioning tradingintegration. See ARCHITECTURE.md,THREAT_MODEL.md, and the release gates inPUBLICATION.md.

Agent-facing tools

Connected MCP clients also receive server-level instructions telling them tocall kernel_status first, distinguish paper from live mode, and never reportan action as successful unless its structured response contains ok: true.

Build and run locally

Prerequisites:

• Go 1.25.13 or newer;
• Node.js 22 only for the official MCP conformance runner; and
• a supported MCP client.

make test
make build
KALSHI_KERNEL_MODE=paper ./bin/kalshi-kernel

Paper mode is the default when KALSHI_KERNEL_MODE is unset. It intentionallydiscards live credential configuration.

Hermes

Add a local stdio server to ~/.hermes/config.yaml:

mcp_servers:
  kalshi-kernel:
    command: "/absolute/path/to/kalshi-kernel/bin/kalshi-kernel"
    env:
      KALSHI_KERNEL_MODE: "paper"

Restart Hermes and call kernel_status before any other tool.

Claude Code

After building the binary:

claude mcp add --scope user --transport stdio kalshi-kernel \
  --env KALSHI_KERNEL_MODE=paper -- /absolute/path/to/bin/kalshi-kernel

Then run claude mcp get kalshi-kernel or open /mcp to confirm theconnection. This local setup is separate from Anthropic's public ConnectorsDirectory, which requires a deployed remote server.

Other stdio clients

Use the standard MCP configuration shape:

{
  "mcpServers": {
    "kalshi-kernel": {
      "command": "/absolute/path/to/bin/kalshi-kernel",
      "env": {
        "KALSHI_KERNEL_MODE": "paper"
      }
    }
  }
}

Execution modes

Paper mode

KALSHI_KERNEL_MODE=paper ./bin/kalshi-kernel

Paper mode will use a local simulated ledger after that backend is implemented.It must never submit an order to Kalshi. Simulated fills will not predict orguarantee live fills, liquidity, latency, slippage, fees, or profitability.

Live mode

Live mode is deliberately difficult to enable:

KALSHI_KERNEL_MODE=live
KALSHI_API_KEY_ID=...
KALSHI_PRIVATE_KEY_PATH=/absolute/path/to/private-key.pem
KALSHI_LIVE_TRADING_ACK=I_UNDERSTAND_THIS_TRADES_REAL_MONEY

These variables currently arm configuration validation only; no live executionadapter is connected. Never commit credentials, private keys, or .env files.Use short-lived credentials where available, a secrets manager in production,and an account or subaccount with the least privileges and capital required.

Specification-driven generation

The curated agent interface lives in specs/mcp-tools.yaml. It maps each MCPtool to exact OpenAPI operation IDs and AsyncAPI channels. Administrative andaccount-management endpoints are not automatically exposed merely because theyexist upstream.

Four authoritative contracts are fetched from https://docs.kalshi.com:

• openapi.yaml — event-contract REST API;
• asyncapi.yaml — event-contract WebSocket API;
• perps_openapi.yaml — perpetuals REST API; and
• perps_asyncapi.yaml — perpetuals WebSocket API.

The fetcher pins HTTPS and the source hostname, limits response sizes, validatesthe expected dialect and non-empty surface, and records hashes and HTTPprovenance. The upstream contract bodies are ignored by Git and fetched into alocal cache as needed. Upstream-derived REST clients are also generated locallyand excluded from source releases. Only the provenance lock, curated MCPoverlay, and generated MCP tool boundary are versioned. Regeneration and testinghappen in a temporary repository copy before a tested lock and MCP boundary canbe promoted.

make upstream-test      # test latest contracts without requiring snapshot parity
make upstream-check     # require parity and run MCP conformance
make upstream-promote   # promote only the exact artifacts that passed

Every CI run performs the strict upstream check; CI also runs daily so drift isdetected even when the repository is idle.

[!WARNING]The downloaded Kalshi contract files identify at least part of the upstreammaterial as proprietary. They are not covered by this project's Apache-2.0license and must not be committed or redistributed. ReviewTHIRD_PARTY_NOTICES.md before publishing generatedartifacts.

Verification

make test              # unit, integration, and contract alignment tests
make test-alignment    # OpenAPI, AsyncAPI, and MCP overlay alignment
make check-generated   # deterministic generated-code check
make test-race         # concurrent calls under Go's race detector
make test-fuzz         # execution-mode parser safety fuzzing
make conformance       # pinned official MCP protocol scenarios
make release-test      # complete local release gate
make public-check      # vulnerability, metadata, and public-readiness gate
make upstream-check    # latest Kalshi contracts plus conformance

The suite checks transport security, authentication contracts, mutationsecurity, operation-ID uniqueness, fixed-point types, curated surface area,required-field parity, strict JSON Schemas, tool titles and safety hints,fail-closed behavior, concurrency, and Streamable HTTP negotiation.

Distribution and registries

The repository contains publication metadata for the official MCP Registry andsubmission dossiers for the OpenAI Plugins Directory and Anthropic directories.Metadata is preparation—not evidence of approval, endorsement, or publication.

Anthropic's current connector review criteria do not accept connectors thattransfer money or other financial assets. Any Claude Connectors Directory buildof this project must therefore be a distinct read-only and/or paper-onlyartifact with live order tools omitted, subject to Anthropic's review.

See PUBLICATION.md for the exact channel matrix, currentblockers, release steps, listing copy, and review test cases. No externalregistry submission is performed automatically.

Contributing and security

Read CONTRIBUTING.md before opening a pull request. Reportvulnerabilities privately according to SECURITY.md; do not open apublic issue for a suspected credential leak or trading-safety vulnerability.General usage help belongs in SUPPORT.md.

Legal and risk notice

This software is provided for development and research. It does not provideinvestment, financial, legal, tax, compliance, or trading advice. Eventcontracts and leveraged products can result in rapid and substantial loss,including loss of the entire amount committed. You are responsible for accounteligibility, jurisdictional restrictions, exchange rules, regulatoryrequirements, taxes, strategy, orders, and losses.

AI systems and software can misunderstand intent, produce incorrect parameters,repeat requests, or behave unexpectedly. Human review, exchange-side risklimits, least-privilege credentials, monitoring, and an independent emergencystop are required before any live deployment.

Read the full DISCLAIMER.md, PRIVACY.md, andSECURITY.md. The project is licensed underApache License 2.0; third-party material is excluded as described inTHIRD_PARTY_NOTICES.md.