Skip to main content
Glama

crypto-mcp

A minimal Model Context Protocol (MCP) server deployed on Vercel for ChatGPT Developer Mode. The baseline functionality (tag v1.0.0) exposes four core tools that allow ChatGPT to search for a symbol, fetch 24h ticker data, retrieve candlestick data, and run a sample dice roll.

Baseline tools

Tool name

Description

search

Parses a symbol query (e.g.

BTCUSDT 1h

) and returns an ID for

fetch

fetch

Returns Binance 24h ticker data for an ID produced by

search

get_binance_klines

Retrieves recent candlestick data for a Binance spot symbol

get_binance_perp_klines

Retrieves USDⓈ-M perpetual (futures) candlestick data

roll_dice

Demonstration tool from the MCP starter template

Advanced tools (price_action_summary, risk policy/order drafting, confirmations, scheduler) are gated behind the environment variable ENABLE_ADVANCED_TOOLS=true. They remain disabled on the public deployment to maximize ChatGPT compatibility.

SSE endpoint

/api/sse returns a lightweight heartbeat stream so that ChatGPT can probe the SSE transport without requiring Redis. Streaming conversations are handled via the Streamable HTTP endpoint (/api/mcp).

Important: ChatGPT validates the presence of search and fetch. Do not remove or rename them; new tools should be added alongside these baseline ones.

Requirements

  • Node.js 20+

  • pnpm 9+

Getting started

pnpm install pnpm dev

The development server listens on http://localhost:3000. You can inspect the MCP endpoint with:

curl -s http://localhost:3000/api/mcp \ -H 'Content-Type: application/json' \ -H 'Accept: application/json, text/event-stream' \ -d '{"jsonrpc":"2.0","id":"1","method":"tools/list","params":{}}'

Local verification

Before opening a pull request run the smoke test which builds the project, starts next start on a random port, and executes the search → fetch → get_binance_klines flow.

pnpm verify:mcp

The CI workflow runs the same script on every push and pull request. The command fails if any of the baseline tools are missing or return an error.

Deploying to Vercel

  1. Push your changes to GitHub.

  2. Ensure the repository is linked to Vercel and that the build command is pnpm install && pnpm build.

  3. After deployment, verify the public endpoint:

    curl -s https://<your-domain>/api/mcp \ -H 'Content-Type: application/json' \ -H 'Accept: application/json, text/event-stream' \ -d '{"jsonrpc":"2.0","id":"1","method":"tools/list","params":{}}'
  4. Add the MCP server in ChatGPT Developer Mode by specifying the full URL https://<your-domain>/api/mcp.

Contributing

See CONTRIBUTING.md for contribution rules, including why the baseline tools and SSE configuration must remain intact.

Release Notes

Historical milestones and release tags are tracked in docs/CHANGELOG.md.

-
security - not tested
-
license - not tested
-
quality - not tested

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/shukehi/crypto-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server