Your AI Agent Wants to Help You Trade. Here's How to Let It — Safely.

---

Your AI agent is not asking to replace your judgment. It is asking for better infrastructure: live market data, cleaner brokerage access, and permission boundaries that let it help without becoming a black box.

This guide is the human side of that setup.

What an Agent Can Actually Do for You

• Check TarsierAlpha's scored candidates every day
• Pull live options chains and account data from Tradier
• Calculate cost, max loss, and trade sizing before you commit
• Monitor open positions and report when stops or targets are near
• Paper trade the workflow before real capital is involved

What the agent cannot do without you:

• Open a brokerage account
• Provide identity documents or your SSN
• Fund the account
• Invent permissions you never granted

Part 1: Open the Tradier Account

Tradier is a brokerage built for API access. That matters because it allows your agent to work from structured account and market data instead of trying to reason from partial information.

• Application time: around 10 to 15 minutes
• Typical approval window: 1 to 2 business days
• Request Level 2 options approval so calls and puts are available

Starting Capital	What It Realistically Supports
$500	Learning the system, one small contract at a time
$1,000 to $2,500	Multiple paper or small live setups with discipline
$2,500 to $5,000	More complete use of the TarsierAlpha workflow

Part 2: Generate the API Token

After approval, visit web.tradier.com/user/api and generate your token.

• Use the Sandbox Token first for paper trading
• Use the Production Token only after the workflow is validated
• Remember: the API token is separate from your account password

Part 3: Connect the Tradier MCP

On Claude, add a custom connector using:

For more advanced CLI-based workflows, use the same MCP endpoint with your API key and keep PAPER_TRADING: true during the testing phase.

Part 4: Connect the Signal Layer

TarsierAlpha is the signal engine. It scans the market, scores setups, and gives your agent a structured starting point for what to analyze next.

A healthy daily workflow looks like this:

1. Agent checks top TarsierAlpha candidates before the open.
2. Agent pulls the live Tradier options chain and account context.
3. Agent presents the best 2 to 3 setups with sizing and max-loss numbers.
4. You approve, reject, or ask for a paper trade version first.

Set the Permissions Clearly

Before your agent can do anything meaningful, define the rules in plain language.

• Require approval for any trade above a fixed dollar threshold
• Require approval for earnings trades or lower-scored setups
• Allow paper trades automatically if they still respect your sizing rules
• Never allow more than 5% of the account in a single position
• Never allow stop losses to be widened after entry

Paper Trading Before Live Trading

This is where trust is earned. Let the agent run the exact workflow in paper mode first.

• Complete at least 10 paper trades
• Review both winners and losers
• Confirm the agent honored stops, sizing, and rules every time
• Move to live trading only after the process is boringly consistent

What “Safe” Actually Means Here

Safe does not mean risk-free. Options trading still carries risk. Safe means the roles are clear:

• The human owns the account, the capital, and the permissions.
• The agent does the monitoring, analysis, and disciplined execution inside those permissions.
• TarsierAlpha provides structured setups rather than hype-driven content.
• Tradier provides the programmable brokerage layer that keeps the workflow auditable.

The Practical Takeaway

If you want an agent to help you trade, start with the boring parts done well: account setup, API access, permission boundaries, and paper-trade validation. That is how trust gets built.