theaisingularity/hermesclaw/market-alerts

market-alerts

This skill monitors your watchlist for price threshold breaches, sends Telegram alerts, and generates daily pre-market briefings. Trade notes and strategy observations are stored in MEMORY.md for future reference.

When to invoke

• On a cron schedule during market hours (every 15 minutes, 9:30am–4pm ET)
• For daily pre-market briefing (9:25am ET weekdays)
• When the user explicitly asks to check their watchlist
• When the user wants to record a trade note

Steps to execute

Mode 1: Real-time monitoring check (cron)

1. Load watchlist and thresholds from memory

• Read MEMORY.md for: ticker symbols, alert thresholds (high/low per ticker), alert preferences
• If no watchlist is configured, ask the user to set one before proceeding

1. Fetch current prices

• Use web_extract to get current prices for each ticker
• Preferred sources: finance.yahoo.com, Google Finance, or similar public price feeds
• Use scripts/monitor.py to parse price data and compute threshold status

1. Check thresholds

• For each ticker: compare current price to high and low thresholds
• A breach = price is above high threshold OR below low threshold

1. Send Telegram alert on breach

• Alert format: see Output Format section below
• Include: ticker, current price, threshold breached, % from threshold, relevant context from MEMORY.md (e.g., "your thesis was X", "last time this happened Y")
• Do NOT send duplicate alerts for the same breach within 60 minutes (track in memory)

1. Update memory

• Record last check timestamp and prices in MEMORY.md
• Track breach history

Mode 2: Pre-market briefing

1. Fetch pre-market prices for all watchlist tickers
2. Fetch overnight/morning news relevant to each ticker (via web_search)
3. Check MEMORY.md for any open position notes or pending thesis items
4. Generate structured briefing (see Output Format)
5. Send to Telegram

Mode 3: Record trade note

When the user says "record trade" or "save my thesis":

1. Extract: ticker, direction (long/short), entry price, thesis, target, stop
2. Save to MEMORY.md under "Trade journal"
3. Confirm to user

Output format

Threshold breach alert:

ALERT — [TICKER] [direction] breach
[TIME]
 
[TICKER]: $[PRICE] ([+/-]%% from threshold)
Your [high/low] threshold: $[THRESHOLD]
 
From your notes: [relevant trade context from MEMORY.md, if any]
 
RSI/momentum context: [brief technical note from web data if available]

Pre-market briefing:

Pre-market — [DATE] [TIME]
 
[TICKER]: $[PRE-MKT PRICE] ([+/-]%% vs yesterday)
  Context: [1-line news or catalyst]
 
[TICKER]: ...
 
Open positions from your notes:
• [TICKER] long entry $[X] — [brief thesis]
 
Today's macro: [1-2 sentences on scheduled events: FOMC, jobs data, earnings]

Trade note confirmation:

Trade recorded:
[TICKER] [LONG/SHORT] | Entry: $[X] | Target: $[Y] | Stop: $[Z]
Thesis: [summary]

Error handling

• If price fetch fails for a ticker: skip it, note in output, retry next cycle
• If Telegram delivery fails: save alert to /sandbox/alert-[timestamp].txt
• If all price sources are unreachable: log the failure and stop the cron check gracefully

Notes

• This skill is for monitoring and analysis only — it does not place orders
• Never store API keys or brokerage credentials in memory
• The scripts/monitor.py helper handles price parsing and threshold comparison as JSON — no fragile string parsing in the LLM
• After 3 months of use, Hermes will auto-create a personalized MarketAlert skill that knows your trading style and typical reaction patterns