priceLABSCO SUMMARY
Market questions answered in the words you would ask them.
It turns "how is the market doing today" and "what is my portfolio worth" into a single live lookup, so an assistant answers with current numbers instead of whatever it remembers about the market.
What it is
A natural-language front end to CoinGecko's public market data, built on FastMCP. It runs on your own machine and every answer is a live lookup — prices, market caps, dominance and 24-hour moves as they stand when you ask.
What you get
- Global market state in one line: total market cap, BTC and ETH dominance, and how much of it moved in 24 hours
- The top coins by market cap with full stats, so a shortlist comes back ranked rather than as a wall of tickers
- The dollar value of holdings you describe, with the best and worst performer in that set called out
- Current price, market cap, volume and 24-hour change for any coin you name
- Two or more coins side by side when the question is the comparison rather than the individual numbers
Requirements
Python with uv or pip, or the Docker image. Market data comes from CoinGecko's free API, which needs no key — a free CoinGecko demo key only matters once you reach the rate limit on the free tier. Every tool call also carries a CryptoSense key, which you mint yourself with the script the repo ships; keys live in a local SQLite file next to the server.
Setup effort
Build from source — git clone https://github.com/josephibra/cryptosense-mcp.git
USE CASES
01
Taking the temperature of the whole market
Ask for the market as a whole and get total market cap, BTC and ETH dominance and the 24-hour move in one answer — the context you want before looking at any single coin.
02
Pricing what you already hold
Describe your holdings and get the dollar value back, with the best and worst performer in the set named. A portfolio check without a spreadsheet.
03
Building a shortlist by size
Ask for the top coins by market cap and get them ranked with full stats, so the list you work from is today's rather than last quarter's.
04
Reading a rally against the rest of the market
Take the top of the market and the global picture together — a coin's move means something different when everything is up.
6 tools
Listed in project docs. Your client may show a different list.
trendingmarket_overviewtop_coinscompareportfolio_valueBEFORE YOU INSTALL
What to know first
Tools you get
6 tools named in the project's documentation.
Requires
python on PATH
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
git clone https://github.com/josephibra/cryptosense-mcp.git
cd cryptosense-mcp
pip install -e .
cp .env.example .env
python generate_key.py <YOUR_EMAIL>Keep the printed cs_… key: every tool call needs it as the api_key argument. The key is stored in keys.db in the folder you ran this from; the client configuration below points the server at that file with DATABASE_URL.
MCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"cryptosense": {
"command": "python",
"args": [
"-m",
"cryptosense.server"
],
"cwd": "<YOUR_PATH_TO_CRYPTOSENSE_MCP>",
"env": {
"DATABASE_URL": "<YOUR_PATH_TO_CRYPTOSENSE_MCP>/keys.db",
"MCP_HOST": "127.0.0.1",
"MCP_PORT": "8000"
}
}
}
}<YOUR_PATH_TO_CRYPTOSENSE_MCP> is the full path of your clone in both places. DATABASE_URL points the server at the keys.db that generate_key.py wrote there; without it the server looks for keys.db in whatever folder the client starts it from, and Claude Desktop does not reliably honour cwd.
