Demerzels-lab Simmer Signal Sniper
Your signals, Simmer's trading intelligence.
When to Use This Skill
Use this skill when the user wants to:
- Monitor RSS feeds for trading opportunities
- Trade on breaking news before markets react
- Configure their own signal sources and keywords
- Get Trading Agent-grade safeguards on their trades
Quick Commands
# Check account balance and positions
python scripts/status.py
# Detailed position list
python scripts/status.py --positions
API Reference:
- Base URL:
https://api.simmer.markets - Auth:
Authorization: Bearer $SIMMER_API_KEY - Portfolio:
GET /api/sdk/portfolio - Positions:
GET /api/sdk/positions
Quick Start (Ad-Hoc Usage)
User provides RSS feed and market directly:
User: "Watch this RSS feed for greenland news: https://news.google.com/rss/search?q=greenland"
User: "Snipe any news about trump from this feed"
→ Run with --feed and --market flags:
python signal_sniper.py --feed "https://news.google.com/rss/search?q=greenland" --market "greenland-acquisition" --dry-run
Persistent Setup (Optional)
For automated recurring scans, configure via environment:
| Setting | Environment Variable | Default | Description |
|---|---|---|---|
| RSS Feeds | SIMMER_SNIPER_FEEDS |
(none) | Comma-separated RSS URLs |
| Markets | SIMMER_SNIPER_MARKETS |
(none) | Comma-separated market IDs |
| Keywords | SIMMER_SNIPER_KEYWORDS |
(none) | Comma-separated keywords to match |
| Confidence | SIMMER_SNIPER_CONFIDENCE |
0.7 | Min confidence to trade (0.0-1.0) |
| Max USD | SIMMER_SNIPER_MAX_USD |
25 | Max per trade |
| Max trades/run | SIMMER_SNIPER_MAX_TRADES |
5 | Maximum trades per scan cycle |
Polymarket Constraints:
- Minimum 5 shares per order
- Trades below this threshold are rejected with an error message
How It Works
Each cycle the script:
- Polls configured RSS feeds
- Filters articles by keywords (if configured)
- Matches articles to target markets
- For each match, calls SDK context endpoint for safeguards:
- Position awareness (already holding?)
- Flip-flop detection (recently changed direction?)
- Slippage estimates (is market liquid?)
- Time decay (resolving soon?)
- Resolution criteria (what actually resolves this market?)
- If safeguards pass, you (Claude) analyze the signal
- If confident, executes trade via SDK
- Tracks processed articles to avoid duplicates
Running the Skill
Run a scan:
python signal_sniper.py
Dry run (no actual trades):
python signal_sniper.py --dry-run
Check for signals without trading:
python signal_sniper.py --scan-only
View current config:
python signal_sniper.py --config
Override for one run:
python signal_sniper.py --feed "https://..." --keywords "trump,greenland" --market "abc123"
Show processed articles:
python signal_sniper.py --history
Interpreting Context Warnings
Before trading, ALWAYS check the context warnings. The skill will show you:
| Warning | Action |
|---|---|
MARKET RESOLVED |
Do NOT trade |
HIGH URGENCY: Resolves in Xh |
Consider if signal is timely enough |
flip_flop_warning: SEVERE |
Skip - you've been reversing too much |
flip_flop_warning: CAUTION |
Proceed carefully, need strong signal |
Wide spread (X%) |
Reduce position size or skip |
Simmer AI signal: X% more bullish/bearish |
Consider Simmer's oracle opinion |
Analyzing Signals
When you find a matching article, analyze it carefully:
Read the headline and summary - What is the actual news?
Check resolution_criteria - What ACTUALLY resolves this market?
- Example: "greenland" in headline doesn't mean "acquisition complete"
- The resolution might be "US formally acquires Greenland by 2027"
- Does this signal move the needle on THAT specific criteria?
Assess confidence (0.0-1.0):
- How directly does this signal relate to resolution criteria?
- Is the source credible?
- Is this news likely already priced in?
Only trade if:
- Confidence > threshold (default 0.7)
- No severe warnings
- Signal validates against resolution criteria
Example Conversations
User: "Set up news sniping for the Greenland market" → Ask for RSS feeds they want to monitor → Configure with market ID and keywords → Enable cron for recurring scans
User: "Check this feed for trading signals"
→ Run: python signal_sniper.py --feed "URL" --scan-only
→ Show found articles and potential matches
User: "Snipe any bitcoin news from CoinDesk" → Run with CoinDesk RSS and bitcoin-related markets → Show matches and ask if they want to trade
User: "What signals have we processed?"
→ Run: python signal_sniper.py --history
→ Show recent articles and actions taken
Example Trade Flow
1. RSS poll finds: "Trump and Denmark reach preliminary Greenland agreement"
2. Keywords match: "greenland", "trump"
3. Call context endpoint for market "greenland-acquisition-2027"
4. Check warnings: none severe ✓
5. Resolution criteria: "Resolves YES if US formally acquires Greenland by 2027"
6. You analyze: "preliminary agreement" ≠ "formally acquires" but bullish signal
7. Confidence: 0.75 (positive indicator, not definitive)
8. Check slippage: 2.5% on $25 ✓
9. Execute: BUY YES $25
10. Report: "🎯 Sniped: Trump/Greenland agreement → BUY YES $25"
Troubleshooting
"No feeds configured"
- Provide feeds in message: "watch this RSS: https://..."
- Or set
SIMMER_SNIPER_FEEDSenvironment variable
"No matching articles found"
- Check keywords are correct
- RSS feed might not have recent articles
- Try
--scan-onlyto see what's in the feed
"Skipped due to flip-flop warning"
- You've been changing direction too much on this market
- Wait before trading again, or find new information
"Slippage too high"
- Market is illiquid
- Reduce trade size or skip
"Already processed"
- This article was already seen
- Working as intended (dedup)
Finding Good RSS Feeds
Tips for choosing signal sources:
- Google News RSS:
https://news.google.com/rss/search?q=YOUR_TOPIC - Niche sources: Better than mainstream (less priced in)
- Official sources: Government, company announcements
- Twitter lists → RSS: Use services like Nitter or RSS.app
The skill works best when:
- Feeds are relevant to your target markets
- You have specific keywords to filter noise
- Sources publish before mainstream coverage