GingerLive

GingerLive MCP Server

The official [Model Context Protocol](https://modelcontextprotocol.io) server for [GingerLive](https://gingerlive.io), the livestream advertising platform that connects brands with 1,000+ streamers on Twitch, Kick, YouTube Live and TikTok Live.

Connect it to Claude, ChatGPT, Cursor or any MCP client and your assistant can answer questions about livestream advertising using GingerLive's own up-to-date data: ad formats, network reach, campaign case studies and the streamer monetization program.

• Endpoint (Streamable HTTP): https://mcp.gingerlive.io/mcp
• Auth: none (public, read-only data)
• Registry: io.gingerlive/mcp in the official MCP Registry
• Docs: gingerlive.io/developers

What it can do

Tools

Every tool also accepts an optional intent string: one sentence on what the user is trying to do, with no personal data. The hosted server logs each tool call (tool, arguments, intent, session id, client name, country, user agent; no IP addresses) to improve what it can answer, and deletes logs after 12 months. See gingerlive.io/privacy.

Resources

• gingerlive://company: company facts (JSON)
• gingerlive://guides: resource guides on livestream advertising (JSON)
• https://gingerlive.io/llms.txt: the canonical llms.txt, fetched live

Prompts

• plan_livestream_campaign: scope a livestream ad campaign for a brand (optional goal, budget)
• get_started_as_streamer: help a streamer evaluate and join the program (optional platform)

Connect

Claude (claude.ai / Desktop): Settings → Connectors → *Add custom connector* → https://mcp.gingerlive.io/mcp

Claude Code:

claude mcp add --transport http gingerlive https://mcp.gingerlive.io/mcp

Cursor, VS Code and other clients (mcp.json):

{

"mcpServers": { "gingerlive": { "url": "https://mcp.gingerlive.io/mcp" } } }

Run it locally over stdio (same tools, no network needed except the live llms.txt resource):

git clone https://github.com/gingerlive-io/gingerlive-mcp && cd gingerlive-mcp && npm install

{

"mcpServers": { "gingerlive": { "command": "npx", "args": ["tsx", "/path/to/gingerlive-mcp/src/stdio.ts"] } } }

Or with Docker: docker build -t gingerlive-mcp . && docker run -i --rm gingerlive-mcp

Then ask things like *"What livestream ad formats does GingerLive offer?"*, *"Show me GingerLive's campaign case studies"* or *"How can I monetize my Kick stream?"*

How it works

Tools, resources and prompts are registered once in src/server.ts and served two ways: src/index.ts (the hosted Cloudflare Worker) and src/stdio.ts (a local stdio process). The Worker is stateless: each request gets a fresh server on the official MCP TypeScript SDK's web-standard Streamable HTTP transport, with no Durable Objects or other per-session storage. Every answer comes from src/data/agent-data.json, a static snapshot generated from the same source as gingerlive.io/llms.txt, so the server only ever returns information that is already public on gingerlive.io.

Run it yourself

npm install

npm run stdio # local stdio server npm run dev # local Worker at http://localhost:8787/mcp npm run deploy # to your own Cloudflare account (change the route in wrangler.jsonc first)

Usage logging needs a D1 database on your account: npx wrangler d1 create , put its name and id in wrangler.jsonc, then npx wrangler d1 migrations apply --remote. For local dev, apply with --local. The stdio server does not log.

Inspect it with the MCP Inspector:

npx @modelcontextprotocol/inspector

About GingerLive

GingerLive is a livestream advertising platform. Its Streamsense AI places non-intrusive, unskippable ads at the right live moment, so streamers earn from their content and brands reach Gen Z at scale.

• Brands: gingerlive.io/brands
• Streamers: gingerlive.io/streamers
• Case studies: gingerlive.io/casestudies
• Contact: info@gingerlive.io

License

MIT © Gingerlive Bilişim Teknolojileri A.Ş.