---
title: "How Do I Connect Trade Inc MCP, and Which Research Tools Can I Use?"
description: "Connect your AI workspace to Trade Inc, sign in, and start researching IDX stocks. A setup guide with available tools, example prompts, Screeners and historical-test limits."
author: "Trade Inc Research"
published_at: "2026-09-30T10:06:24Z"
updated_at: "2026-09-30T10:15:09Z"
canonical: "https://tradeinc.app/en/blog/idx-research-existing-ai-tool-mcp"
tickers: []
citations: ["https://tradeinc.app/en","https://modelcontextprotocol.io/docs/learn/architecture"]
---

# How Do I Connect Trade Inc MCP, and Which Research Tools Can I Use?

Connect your AI workspace to Trade Inc, sign in, and start researching IDX stocks. A setup guide with available tools, example prompts, Screeners and historical-test limits.

## Summary

Connect Trade Inc to an AI application that supports remote MCP over Streamable HTTP, sign in with your Trade Inc account, and ask it to use Trade Inc’s research tools. You can retrieve IDX prices, financials, disclosures, broker activity and ownership evidence, discover Screeners, and work with historical Screener tests. Available tools depend on your account access. [Trade Inc connection information](https://tradeinc.app/en)

This guide covers the connection settings, the research-tool catalogue and prompts to try after connecting. Updated 30 September 2026. Published by Trade Inc Research, affiliated with Trade Inc.

## Key takeaways

- Server URL: `https://tradeinc.app/api/mcp`.
- Connection type: remote MCP using **Streamable HTTP**.
- Authentication: sign in with your Trade Inc account through the connection flow.
- Start with a read-only ticker lookup. Saving a Screener or starting a historical test is a separate action.

## How to connect Trade Inc MCP

### 1. Check your AI application’s connection support

Your application must support adding a remote MCP server and completing its account-authentication flow. Look for its integrations, connectors or MCP-server settings; the exact menu name depends on the application and version. A chat box alone is not a server-configuration screen.

MCP is the connection standard, not a separate AI model. It allows an AI application to discover and call tools supplied by a server. [Official MCP architecture](https://modelcontextprotocol.io/docs/learn/architecture)

### 2. Add the server

Create a remote-server connection using these values:

| Setting | Value |
| --- | --- |
| Display name | Trade Inc |
| Server URL | `https://tradeinc.app/api/mcp` |
| Transport, if requested | Streamable HTTP |
| Account | Your Trade Inc account |

“Trade Inc” is a suggested display name. Use the full server URL, including `/api/mcp`, rather than the website homepage. This is a remote connection; these instructions do not require a local server command.

### 3. Sign in and review access

Follow the sign-in prompt and review the access request before accepting it. Return to your AI application when authentication finishes. Do not paste your password or an access token into a conversation.

The endpoint, transport and account sign-in are documented in the [Trade Inc FAQ](https://tradeinc.app/en). Client-specific buttons and account entitlements can differ; this guide does not promise that every AI client or plan supports the connection.

### 4. Confirm the tools are available

Check the connection’s tool list, if your application displays one. Then send:

> Use Trade Inc to look up BBCA. Show the company name, latest available daily close and the date of that close. Do not create or run a Screener or backtest.

The ticker lookup uses `get_ticker`. The response should identify the data date, not merely the time you asked. If the assistant answers without using the connection, explicitly select or enable Trade Inc using your client’s controls and retry.

## Which research tools are available?

The following catalogue describes the research tools exposed by the connected service when this guide was updated. Your authenticated tool list determines what your account can actually use. These are MCP capabilities, not a promise that every function has an identical button in the website.

### Company, price and evidence tools

| Tool | What it does | Example request |
| --- | --- | --- |
| `get_ticker` | Resolves an IDX ticker and its latest daily close | “Look up BBCA and show the close date.” |
| `get_candles` | Retrieves completed daily or weekly open, high, low, close and volume data, with named candlestick-pattern checks | “Show TLKM’s completed daily candles over the last 30 days and flag doji or hammer patterns.” |
| `get_financial_series` | Retrieves annual and quarterly financial figures and ratios | “Show ASII’s last four available quarters and state the reporting periods.” |
| `get_disclosure_evidence` | Retrieves official IDX disclosure metadata and source links | “Find BBCA disclosures from the last 30 days and link the original sources.” |
| `get_market_events` | Retrieves recent company news and corporate actions, for a ticker or the market | “Summarize recent TLKM news and corporate actions with dates.” |
| `get_broker_activity` | Retrieves broker accumulation or distribution estimates for a ticker over a bounded window | “Show BBCA broker activity over the last 10 days; separate purchases, sales and net activity where available.” |
| `get_ownership_timeline` | Retrieves ownership filings and register history, with broker-flow context kept separate | “Show ASII’s ownership timeline and distinguish filings from broker activity.” |
| `get_market_regime` | Retrieves stored breadth, trend mix and foreign-flow context for completed IDX sessions | “Summarize the last five completed sessions’ market context.” |

Candlestick labels describe shapes, not forecasts. Broker activity is not proof of a beneficial owner. A disclosure listing may point to a document that still needs to be opened and read.

### Market discovery and Screeners

| Capability | What you can ask |
| --- | --- |
| Current market ranking — `screen_market` | “Show 10 names from the current Screener ranking and explain the returned price, liquidity and broker metrics.” |
| Published official calls — `list_calls` | “List published swing calls and their latest tracked state.” |
| Official Screener catalogue | “List the available official Screeners and their latest run status.” |
| Your accessible Screeners | “List my saved and accessible official Screeners so I can choose one.” |
| Save a new Screener | “Help me define a Screener. Show the proposed rules and schedule before I approve saving it.” |

The service supports creating a user-owned Screener after an explicit save request. That is different from displaying the current market ranking. Review the proposed definition before saving; do not assume an example prompt is a fully specified strategy.

### Historical tests and reports

| Tool | What it does |
| --- | --- |
| `create_backtest` | Starts a historical, point-in-time rule replay for an accessible saved Screener, using a specified start and end date |
| `get_backtest_report` | Retrieves an existing report using the returned run or job identifier; it does not start another test |
| `compare_strategies` | Compares two to four stored runs or completed replay jobs without rerunning them |

For a new test, select the exact saved Screener and provide both dates. For example:

> Show my accessible Screeners first. After I choose one, I want to test it from 2026-08-01 through 2026-09-16. Confirm the selected Screener and dates before starting. Keep the returned identifier so we can retrieve the report.

The current MCP creation tool reproduces historical Screener matches. It does **not** simulate trade fills or trading returns. A submitted job is also not a completed report: wait for completion, then retrieve the result. For comparisons, use identifiers from the same supported result family rather than mixing stored-run and replay-job identifiers.

## A useful first research conversation

After the ticker lookup works, build one bounded investigation:

1. “Use Trade Inc to retrieve BBCA’s latest available financial periods and recent disclosure links.”
2. “Separate facts from interpretation. Show the reporting periods and flag missing figures.”
3. “Add recent broker activity as a separate section; do not treat it as ownership proof.”

These are example prompts, not precomputed results or trade recommendations. They combine retrieval tools without creating saved research or launching a test.

For more detail, see our [financial-statement guide](https://tradeinc.app/en/blog/ai-tool-indonesian-financial-statement-analysis) and [broker-analysis guide](https://tradeinc.app/en/blog/ai-broker-analysis-indonesian-stocks).

## If the connection does not work

| Symptom | What to check |
| --- | --- |
| The client asks for a local command | Check whether you selected a local-server configuration instead of a remote URL connection |
| Sign-in does not finish | Retry the connection’s sign-in flow and confirm you are using the intended Trade Inc account |
| Connected, but a tool is missing | Refresh the client’s available tools if supported; check account access rather than assuming every capability is included |
| The assistant gives a generic answer | Ask it to use Trade Inc explicitly and inspect whether a tool call occurred |
| A report cannot be found | Confirm that a test was actually started, completed and queried with its returned identifier; do not invent an ID |
| Data looks old or incomplete | Inspect the returned session dates, reporting periods and coverage before drawing a conclusion |

## Conclusion

Add the remote server, authenticate, and verify one ticker lookup before expanding your workflow. Then choose the tools that match your question: company evidence, market context, screening or historical rule evaluation. Keep reading data separate from saving a Screener or starting a test, and check the source dates behind each answer.

Trade Inc MCP brings research tools into a compatible AI workspace; it does not turn a model’s interpretation into investment advice.


[1]: https://tradeinc.app/en "Trade Inc product overview"
[2]: https://modelcontextprotocol.io/docs/learn/architecture "MCP architecture overview"