Multi-Asset Portfolio Watch with One AI Agent
Holdings spread across A-shares, US stocks, crypto and forex? Use one AI agent to pull quotes from every market, convert to a single currency, and produce one portfolio watch report — no app-hopping.

Tool names in this guide are for reference only and may have changed or been renamed. Please use the latest tools in the tool marketplace when implementing.
Key takeaways
- Pull A-share, US-stock, crypto and forex quotes with one agent instead of juggling apps
- Normalize multi-currency quotes to a single base currency and compute portfolio P&L in one pass
- Turn the whole flow into a reusable watch report you can run on a schedule
Tools used
Why cross-market watching is such a chore
Anyone with a slightly diversified portfolio ends up with a pile of apps: one for A-shares, one for US stocks, one for crypto, and yet another tab for the exchange rate. Every source has its own quote format, refresh cadence and currency. To answer a simple question — "is my whole portfolio up or down today?" — you have to jot down each position by hand, convert CNY, USD and crypto holdings into one currency, and add it all up. By the time you finish, the prices have already moved.
The problem isn't access to data. It's that the data lives in different markets, different APIs and different currencies, with nothing orchestrating it for you. A single quote API usually covers just one asset class, so the cross-market roll-up is left to you.
The workflow in five steps
Break "see the whole portfolio" into one reproducible loop:
- Hand the agent a holdings list spanning A-shares, US stocks, crypto and forex.
- The agent pulls quotes from each market by asset class.
- Convert every currency to one base currency using live exchange rates.
- Compute per-position and whole-portfolio moves against your holdings.
- Roll it all up into one consistently formatted watch report.
The key: all five steps are done by one agent calling several data sources, not by you shuttling between five apps.
How each step works
Step 1: Hand the agent your holdings
Describe your positions in natural language, e.g. use ae to watch my portfolio: 1,000 shares of A-share 600000, 50 shares of AAPL, 0.5 BTC, plus a USD cash balance — summarize it in CNY. You don't pick a tool; the agent routes by asset class.
Step 2: Pull quotes per market
The agent sends each asset to its category tool: an A-share daily-quote tool (such as ae_tushare_daily), a US-stock quote tool (such as ae_twelvedata_get_quote), and a crypto spot-price tool (such as ae_coingecko_simple_price). All in one conversation, in parallel.
| Asset class | Tool category | What you get |
|---|---|---|
| A-shares | Daily-quote (e.g. ae_tushare_daily) |
Close / change |
| US stocks | Equity quote (e.g. ae_twelvedata_get_quote) |
Live quote |
| Crypto | Digital-asset spot (e.g. ae_coingecko_simple_price) |
Coin price |
| Forex | Exchange rate (e.g. ae_twelvedata_get_exchange_rate) |
Currency pair |
Step 3: Normalize currencies
A portfolio mixing CNY, USD and crypto can't just be summed. The agent calls an exchange-rate tool (such as ae_twelvedata_get_exchange_rate) for pairs like USD/CNY and converts every position into the base currency you named.
Step 4: Compute portfolio moves
With quotes in a common currency, the agent uses your position sizes to compute each position's market value and daily change, then totals the portfolio's P&L and percentage move.
Step 5: Produce the watch report
Finally the agent assembles a structured report: per-position value, daily change, total portfolio value and percentage move. Want it at the same time every day? Hand this instruction to an agent flow that can run on a schedule.
FAQ
Q: Which markets are supported? A: The example covers A-shares, US stocks, crypto and forex. The agent picks a category tool per asset class, and the same AgentEarth connection reaches 2,400+ other tools.
Q: Can I really see everything in one view? A: Yes. The whole point is "one agent, many sources" — cross-market quotes get normalized and rolled up into a single report, with no app-switching.
Q: Can it run on a schedule? A: It can. Save this watch instruction and hand it to an agent flow that supports scheduled triggers to produce the report at a fixed cadence.
How to get started
- Sign up — new users get 1,000 free credits.
- Connect AgentEarth to your agent — in-app search in WorkBuddy / Trae, or an MCP config line for Claude Code / Cursor / VSCode.
- Give the instruction in natural language and let the agent run the whole watch flow for you.
