Kensho Sweep Mcp

Read-only MCP server that lets an assistant search a bundled dataset of 1133 Japanese X/Twitter sweepstakes campaigns and track prize value changes.

What it does

Exposes a local snapshot of Japanese X/Twitter sweepstakes campaign data collected from several campaign listing sites. Three tools are provided: current_sweep(keyword) returns the latest sweep matching a keyword such as a prize, brand, or URL substring; sweep_history(keyword, limit=50) returns a price time series oldest first; top_prize_movers(direction=None, limit=10) ranks the biggest absolute JPY prize value movers with an optional up/down filter.

When to use it

Useful when an assistant needs to look up Japanese sweepstakes campaigns, check a campaign's prize history, or see which prizes have changed most in value, without browsing the web. It is a read-only reference dataset, so it suits research and comparison questions rather than live campaign tracking.

Requirements

Runs as a local process over stdio (or streamable-http at /mcp with the --http flag). Needs Python and the fastmcp package version 3.0.0 or later. No network access, API key, or account is required; the data is bundled in the package.