Skip to main content

TradingFlow AI

The same options data.
For you,
and for your agent.

Traders run Rank → tape → structure. Agents query the same data through MCP. Default access is read-only. Execution, if it happens, is a separate user gate on a connected broker. We do not trade for you.

Paid access. AI Chat, recipe editing, and MCP availability depend on your account.

Example workflow

“What stands out in this options flow?”
Your current view
Options flow
Market structure

An explanation you can investigate.

Ask a follow-up, inspect the underlying rows, or request a chart.
Explore AI Chat

AI Chat

A research partner.
Right beside the data.

Keep your question next to the chart, ranking, or tape that prompted it. TradingFlow AI understands the view you have open and can investigate with market-data tools.

  • Explain the flow, positioning, or GEX you’re looking at.
  • Go deeper with follow-up questions and focused page context.
  • Get visual answers with charts, tables, and summary metrics.
Explore AI Chat
TradingFlow AI docked beside Rank Symbols, with the current page attached as context and a research question composer.
Test app · AI sidebar on Rank Symbols · September 8, 2026. Captured session data, not a live feed.
Try asking

“Compare the strongest net-premium names in this view. What should I inspect on the tape next?”

Recipes

Build it once.
Research it again.

Turn the questions you ask every session into a report you can rerun. Start with an official Cookbook or describe your own idea, then refine its charts, tables, and inputs with AI.

  • Run reports for the latest completed session or a date you choose.
  • Use AI Insight to explain the results of a specific run.
  • Preview edits and explicitly save your own private recipe.
Explore Cookbooks
Daily Market Recap in Cookbooks, with a session picker, Run, AI Insight, Edit with AI, and data-driven report chapters.
Test app · Daily Market Recap recipe · September 8, 2026. Reports identify the session represented.
DescribePreviewSaveRun again
Try building

“Create a net-premium leaders report with a bar chart, a table, and a symbol filter.”

MCP

Your agent.
Connected to live options flow.

Add one remote MCP server and Claude, Cursor, Codex, or OpenClaw queries the same Rank, tape, and structure data you use in the app. The default connection is read-only. It cannot send an order.

Connect an agent

MCP server URL

https://app.tradingflow.com/api/mcp

Add this URL in your client’s MCP settings. Claude signs in with your TradingFlow account. Cursor, Codex, OpenClaw, and scripts send an API key from the Connect page.

Or paste this into your coding agent and it connects itself:

Connect this project to TradingFlow's MCP server at https://app.tradingflow.com/api/mcp (streamable HTTP). Claude Code signs in: run `claude mcp add --transport http tradingflow https://app.tradingflow.com/api/mcp`, then `claude mcp login tradingflow`. Any other client sends an x-api-key header with a key from https://app.tradingflow.com/app/settings/mcp-connect; ask me for the key, never invent one. The server is read-only and cannot send an order. Read the server's tool list before calling anything; do not invent tool names. Reference: https://app.tradingflow.com/docs/mcp

Read the live tape

Option prints, aggregate flow, candles, and contract rankings through eight read-only data tools.

Query governed metrics

Named metrics and dimensions compiled to SQL. Your agent reads the catalog first, then asks in those terms.

Stay in control

Read-only by default. Access is checked on every call, and recipe writes need a separate opt-in key.

Paid account with MCP access required. Your agent’s own usage charges apply; MCP queries do not use TradingFlow AI credits. Claude connects by signing in; any other client sends an API-key header.

Start with the session. Then go deeper.

Daily Market Recap brings session tone, activity, exposure, volatility, and market structure into one Cookbook report. Choose a session, run the report, and ask AI Insight to help you read it.

Open Market Recap

Before you
start asking.

How the tools work, what you can connect, and where you stay in control.

View plans

What can I ask TradingFlow AI?

Ask it to explain the Rank view, investigate options flow, compare a symbol’s positioning, or visualize a market-data question. The assistant uses your current page context and read-only tools. It can answer with text, charts, tables, and links to the relevant app views. Review its output against the underlying data.

What is a recipe?

A recipe is a reusable research report with data queries, charts, tables, and summary blocks. Open an official template in Cookbooks, choose a session date, and run it. When recipe editing is enabled for your account, you can create a private recipe or fork a template, refine it with AI, review the preview, and explicitly save it under My recipes.

Which agents can connect through MCP?

Claude signs in with your TradingFlow account: claude.ai, Claude Desktop, the mobile apps, Cowork, and Claude Code connect with no API key. Cursor, Codex, OpenClaw, and any other client that sends an API-key header connect with a key from the in-app Connect page. ChatGPT and other hosted chat sites that cannot send a key are not supported. The setup card above has each client’s current snippet.

How do access and AI credits work?

Cookbooks requires a paid subscription. AI Chat, recipe authoring, and MCP also depend on account availability. In-app AI replies require consent and use TradingFlow AI credits. MCP data queries do not use TradingFlow AI credits; reasoning is handled by your connected agent and uses that agent’s own plan or API budget.

Will the AI save changes or place trades for me?

TradingFlow is a research tool and does not place trades. In-app AI recipe changes are reviewed in a preview and saved explicitly by you. MCP uses read-only access by default. Optional recipe-write permissions are separate and require explicit setup and confirmation.

Is every number live?

Different datasets have different clocks. Options flow can update during the session, while open interest and some market-structure metrics use dated snapshots. Check the represented session and freshness labels. Recipes run for the session you select; their results and AI explanations are research inputs, not predictions or investment advice.

Follow your next question.

Explore the flow, build a repeatable report, or connect the agent you already work with.

Research tools for self-directed traders. AI can make mistakes; verify the data and apply your own risk controls.