by lorine93s
Polymarket AI market suggestor that blends real-time news, social sentiment, and on-chain trends to propose new prediction markets with LLM reasoning, analytics, and CLI tooling.
# Add to your Claude Code skills
git clone https://github.com/lorine93s/polymarket-ai-market-suggestorAI-native workflow that transforms real-time trends, news, and existing Polymarket markets into actionable, publish-ready market suggestions. Built for research teams, traders, DAO governance desks, and community curators who want to discover “what market should Polymarket list next?” using a mix of LLM reasoning, sentiment signals, and overlap checks.
SEO keywords: Polymarket AI, AI agent Polymarket, Polymarket AI market prediction, AI market prediction, Polymarket prediction market bot, Polymarket real-time trading bot, Polymarket AI trading bot, Polymarket market analyzer, Polymarket AI market suggestor.
No comments yet. Be the first to share your thoughts!
ℹ️ The AI core and workflow are inspired by leading open-source prediction market agents such as Prediction-Market-AggregationAgent-system while focusing on generative market ideation instead of trading.
TrendScanner (NewsAPI, Twitter API)
│
▼
Sentiment & Keyword signals ────────┐
│
PolymarketClient (Gamma API) │
│ │
▼ │
Existing market snapshots ──────────┘
│
▼
SuggestionEngine (LangChain + GPT-4o)
│
▼
SuggestionBundle (Pydantic)
│
├─ Storage (SQLite) — durable bundle history + analytics
├─ CLI (Typer + Rich) — interactive reports
└─ Reporting utils — JSON / Markdown / history summaries
trend_scanner.py — pulls hot news and tweets with VADER sentiment.trend_scanner.py (crypto) — optional CoinGecko trending feed for DeFi narratives.polymarket_client.py — fetches trending/current markets via Gamma API.ai.py — formats context + prompts GPT-4o (falls back if no API key).orchestrator.py — runs the end-to-end pipeline and deduplicates output.storage.py — SQLite-backed bundle persistence.analytics.py — portfolio-wide stats on past runs.reporting.py — exports Markdown dashboards and history summaries.cli.py — Typer command group: suggest and summarize.git clone https://github.com/your-org/polymarket-ai-market-suggestor.git
cd polymarket-ai-market-suggestor
python -m venv .venv && source .venv/bin/activate # or use uv/pdm
pip install -r requirements.txt
cp ENV.sample .env
# populate .env with OpenAI / NewsAPI / Twitter keys
polysuggest suggest "AI safety regulation" --keywords "AI,regulation,legislation" --count 4 \
--markdown reports/ai-safety.md --output reports/ai-safety.json
Sample console output:
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Title ┃ Confidence ┃ Resolution Source ┃ Tags ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ Will the EU enact the AI Act before Q4 2025? │ 0.72 │ Official EU Parliament DB │ ai-policy,... │
│ ... │ ... │ ... │ ... │
└───────────────────────────────────────────────┴────────────┴────────────────────────────┴───────────────┘
Outputs:
reports/ai-safety.json – structured SuggestionBundle.reports/ai-safety.md – Markdown one-pager for sharing.data/bundles.db) captures a full copy for analytics.polysuggest summarize reports/
Displays a Rich table of prior suggestion runs (topic, timestamp, top pick, confidence).
To use the built-in storage instead:
polysuggest summarize
polysuggest show 3 # Detailed view for run #3
polysuggest insights # Aggregated stats (top tags, avg confidence, sentiment)
Environment variables (copy ENV.sample):
| Variable | Description | Example |
| --- | --- | --- |
| OPENAI_API_KEY | GPT-4o API key for suggestion engine | sk-... |
| OPENAI_MODEL | Override model | gpt-4o-mini |
| POLYMARKET_API_BASE | Gamma API endpoint | https://gamma-api.polymarket.com |
| NEWS_API_KEY | NewsAPI key (optional) | news-... |
| TWITTER_BEARER_TOKEN | Twitter v2 bearer token (optional) | AAAAAAAA... |
| DEFAULT_TREND_KEYWORDS | Fallback keywords | polymarket, ai, elections |
| CHROMA_PERSIST_PATH | Future use for RAG vector store | .chroma |
| POLYSUGGEST_DATA_DIR | Directory for SQLite bundle storage | data |
No LLM key? The system falls back to a deterministic heuristic generator so pipelines remain testable offline.
pip install -r requirements.txt
pytest
To run the CLI inside the repo without installing:
python -m polysuggest.cli suggest "US election turnout"
Logging is powered by Loguru; set LOGURU_LEVEL=DEBUG for verbose traces.
Dockerfile coming soon (project is fully dependency-pinned via requirements.txt / pyproject.toml).
Let’s build the next generation of AI-native Polymarket tooling together.