Read signals from a specific Twitter/X account + Substack
• Use an LLM to extract BUY/SELL, ticker, strike, expiry
• Run strict safety checks + paper trading only (no real money yet)
Not trying to over-engineer this with heavy agents — keeping it Python + simple LLM parsing + Alpaca paper API for v1.
Curious:
What’s the cleanest architecture you’d use for signal extraction → validation → execution?
Any open-source tools or agent frameworks actually worth using here, or better to stay minimal?
Would love ideas from folks experimenting with autonomous/semiautonomous trading workflows.