AgentPlays
Build a personal investing and home-search research agent
A developer built a multi-agent system with mcp-agent that pulls market data and writes daily reports based on their risk tolerance, then reused it to compare neighborhoods for real estate.
Try it yourself
What you get
Daily research reports matched to your risk tolerance, plus help comparing neighborhoods (flood risk, schools, weather).
Tools
- mcp-agent (open source)
- GPT-4 or local models via Ollama
Steps
- Start from the open-source example in the mcp-agent repo.
- Set your preferences, such as risk tolerance and location needs.
- Let the orchestrator route tasks to agents and the evaluator refine results.
Copy-ready prompt
Compare these 3 neighborhoods for buying a home: [list]. For each, report flood risk, school zones, typical weather and the age of the housing stock, with sources. Then rank them for my priorities: [priorities].
Watch out: Research aid, not financial advice.
Proof / source
Open-source code Author's Show HN post and the public GitHub example.
Read the original source ↗ Discussion ↗
Related
Build a paper-trading bot
User's Grok Bot built a paper trader; a follow-up post says it was up $400 the next day (paper money, not real).
Dispute a charge, order lunch and clear spam in the first hours
In the first couple of hours, the Hark agent disputed a transaction over email (saving a few hundred dollars), ordered lunch via Uber Eats, bought a book on Amazon, set u…
Check whether a photographed bill was already paid
User photographed a bill; Grok Bot searched the household financials and found it had not been paid yet.
Catch a bug in a trading plan before money moved
Grok Bot found a bug in a (paper) trading plan before any fake dollar moved, fixed the plan and reran the test.