Crypto Signals

Crypto Trading Signals

Trading signal generation for a $10 crypto portfolio: $5 BTC, $5 ETH, plus crypto-adjacent stocks COIN and MSTR.

Portfolio

ASSET  | TYPE    | VALUE
-------|---------|--------
BTC    | Crypto  | $5.00
ETH    | Crypto  | $5.00
COIN   | Stock   | Track only
MSTR   | Stock   | Track only

COIN and MSTR are stocks, not cryptocurrencies. They trade Mon-Fri 9:30AM-4PM ET.

How It Works

This skill uses a two-phase design optimized for small models (0.5B-1B):

Phase 1 — Data (script does everything): A Python script fetches prices, computes RSI/MACD/Bollinger indicators, generates BUY/HOLD/SELL signals, and writes the results to crypto_report.json.

Phase 2 — Interpret (model reads JSON, writes summary): The model reads the pre-computed JSON and writes a human-readable report to crypto_summary.txt.

The script handles all the hard work. The model only needs to call 2 tools total.

Workflow

Step 1: Run the signal script

IMPORTANT: Copy and paste this exact command into the shell tool. Do NOT modify the path.

find /usr -name crypto_signal_agent.py -exec python {} run \;

This command searches for the script and runs it. The path will be found automatically. Do NOT guess paths like /path/to/ or /usr/bin/python. Use only the command above, exactly as written.

The script will:

• Fetch BTC and ETH prices from CoinGecko (fallback: CoinCap)
• Fetch COIN and MSTR prices from Yahoo Finance
• Compute RSI, SMA, EMA, MACD, Bollinger Bands from 30-day history
• Generate BUY/HOLD/SELL signals with confidence levels
• Write results to crypto_report.json in the current directory
• Append to crypto_signals_log.json for historical tracking

Step 2: Read the report

Use the read_file tool to read crypto_report.json.

Step 3: Write a human-readable summary

Based on the JSON data, use write_file to create crypto_summary.txt containing:

CRYPTO SIGNAL REPORT — {date}
========================

BTC: ${price} ({change_24h}% 24h)
  RSI(14): {rsi} | MACD: {macd} | BB Position: {pos}
  SIGNAL: {signal} ({confidence})

ETH: ${price} ({change_24h}% 24h)
  RSI(14): {rsi} | MACD: {macd} | BB Position: {pos}
  SIGNAL: {signal} ({confidence})

COIN: ${price} ({change}% 24h) — STOCK, not crypto
MSTR: ${price} ({change}% 24h) — STOCK, not crypto

PORTFOLIO: ${total} | 24h P&L: {pnl}%
ACTION: {one sentence recommendation}

Use the signal data from the JSON. Do not make up numbers.

Step 4: Present to user

Read crypto_summary.txt and display it as your final answer.

Quick Commands

• "Run signals" → Full 4-step workflow above
• "Check prices" → Run script, read report, show prices only
• "How's my portfolio" → Run script, read report, show portfolio section

Other Script Commands

The script supports additional commands:

• run — Full analysis + JSON report
• prices — Current prices only
• history — Last 100 logged signals

Change run to prices or history in the command above.

Constraints

• Zero dependencies — Python stdlib only
• Free APIs only — CoinGecko, CoinCap, Yahoo Finance (no keys)
• COIN/MSTR are stocks — note this in output
• Not financial advice — for educational/research purposes

Troubleshooting

• If the script fails with a network error, the JSON will contain _error fields
• If CoinGecko rate-limits (429), the script auto-falls back to CoinCap
• If Yahoo Finance fails, stock data will be missing (crypto still works)
• The script writes to the current working directory

Asset Reference