bringfour
MCP server that turns AI assistants into VGC teambuilding partners with real usage stats, damage calculations, and threat reports for competitive Pokémon.
README
bringfour
Give your AI assistant a competitive Pokémon brain. An MCP server (plus a local web dashboard) that turns Claude, Cursor, or any MCP client into a VGC teambuilding partner — real usage stats, real damage calcs, and threat reports grounded in math instead of type-chart guessing.
🚧 Pre-1.0: the npm publish is imminent — until it lands, use the from-source setup; everything works today.
<!-- TODO(launch): record docs/threat-report.gif — paste team → threat cards -->

Ask "what beats my trick room team?" and get back OHKO/2HKO lines computed
with @smogon/calc against the top meta Pokémon at their actual most common
spreads — not vibes. Defaults to the current official format (Pokémon
Champions, Regulation Set M-B); SV Regulation I and other formats are one
format parameter away.
Quick start
Claude Code
claude mcp add bringfour -- npx -y bringfour
Claude Desktop / Cursor
{
"mcpServers": {
"bringfour": {
"command": "npx",
"args": ["-y", "bringfour"]
}
}
}
From source
git clone <this repo> && cd <repo> && npm install
The repo ships a project-scoped .mcp.json, so Claude Code picks the server
up automatically when working in this directory. For other clients, run
npx tsx src/index.ts (stdio transport).
Tools
| Tool | What it does |
|---|---|
meta_snapshot |
Top meta Pokemon with usage/win rates for a format. |
pokemon_deep_dive |
One Pokemon's common items, abilities, moves, spreads, natures, Tera types (SV), teammates. |
analyze_team |
Calc-backed threat report: for the top N meta Pokemon at their most common sets, real OHKO/2HKO threats against your actual EVs, plus speed tiers and speed-control flags (Tailwind/Trick Room/Icy Wind...). |
calc_damage |
Damage calc via @smogon/calc. Unspecified fields auto-fill from meta data; every assumption is echoed so the agent can override and re-call. |
save_team / list_teams |
Store Showdown-export pastes as plain text, so "analyze my worlds team" works without re-pasting. |
Web dashboard
npm run ui # http://localhost:4747
A local dashboard (dark, sprite-heavy, no build step) over the same data layer as the MCP tools:
- Threats — pick a saved team, get the calc-backed threat report as cards with OHKO/2HKO badges, damage bars, a speed ladder, and speed-control flags.
- Damage Calc — attacker/defender panels that pre-fill the most common meta set (grayed-italic until you override a field), live results, field conditions, swap sides.
- Teams — create/edit Showdown pastes, or import from a pokepast.es link.
- Meta / deep-dive — sortable usage table; click through to per-Pokemon items, moves, spreads, natures, and teammates.
Where data lives
Running from a checkout (or any directory containing a teams/ folder) keeps
teams in ./teams/ and the 24h usage-data cache in ./.cache/, as always.
Running via npx/global install falls back to ~/.bringfour/. Override with:
| Env var | Effect |
|---|---|
VGC_TOOLS_HOME |
Base directory for both (<home>/teams, <home>/cache) |
VGC_TEAMS_DIR |
Teams directory only |
VGC_CACHE_DIR |
Cache directory only |
Data sources & behavior
- Pikalytics (pikalytics.com) is the primary usage source — in-game ranked battle data for Champions, HOME battle data for SV. Pikalytics asks for attribution when citing its stats; every tool response carries it.
- Smogon usage stats (smogon.com/stats, Showdown ladder chaos JSON) are the automatic fallback when Pikalytics lacks a format.
- Usage dumps are monthly; the service walks back up to 7 months to find the newest published data, and caches responses for 24h.
- Teams are plain Showdown-export text files — edit them freely.
Champions caveats (honest-approximation zone)
- Champions EVs run 0–32 per stat. For damage/stat math they are mapped onto
the classic 0–252 scale (×8, capped); every calc that does this says so in
its
assumptions. Treat very close rolls with caution. - Champions Mega formes ("Charizard-Mega-Y") mostly exist in
@smogon/calc's data already; anything the calc doesn't know is reported explicitly instead of guessed at.
Roadmap
- v2: EV-spread optimization; tournament results/team lists via the documented Limitless API.
- Hosted copilot: a web app where a frontier model writes full team coaching reports over these tools — in design, gated on this repo proving demand. Star the repo if you want it to exist.
Support
<!-- TODO(launch): add GitHub Sponsors / Ko-fi link --> If this saves you a ladder session of scouting, a star is the best signal you can send.
License & legal
MIT — see LICENSE.
Unofficial fan tool — not affiliated with, endorsed, or supported by Nintendo, Creatures Inc., GAME FREAK, or The Pokémon Company. Pokémon and all respective names are trademarks of their owners. Usage data courtesy of Pikalytics (attribution requested) and Smogon usage stats; damage math by @smogon/calc (MIT); sprites served from Pokémon Showdown's public sprite sets.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.