# ThinkingData MCP

> ThinkingData / 数数科技 MCP connects your AI agent directly to a major gaming analytics platform. It lets you audit complex user behavior, query event data across massive datasets, and manage player profiles without ever touching a dashboard or writing SQL. Turn deep-dive data analysis—like tracking retention funnels or auditing live-ops events—into simple conversation with your agent.

## Overview
- **Category:** data-analytics
- **Price:** Free
- **Tags:** behavioral-analytics, event-tracking, user-profiling, cohort-analysis, gaming-insights, data-querying

## Description

ThinkingData gives your AI agent the power to run complex analytics queries normally reserved for dedicated data teams. You stop navigating confusing dashboards and start talking to your data. Your agent can instantly pull project metadata, list all defined events in a game, or execute behavioral queries that map out exactly how players are moving through your app. If you're running a player retention audit, the agent pulls the necessary event sequences; if you need to track custom actions for testing, it handles those data inputs too. This means coordinating live-ops refreshes or understanding user funnels becomes a natural conversation. With Vinkius, connecting ThinkingData is straightforward, giving your AI client access to professional-grade insights from one single source.

## Tools

### get_event_properties
Retrieves specific property details for a given event.

### get_project_summary
Provides an overall status and summary of the current data project setup.

### list_defined_events
Retrieves a list of all event types tracked by the system.

### list_project_cohorts
Lists predefined groups or cohorts of users for analysis.

### query_events
Runs a comprehensive query against historical event data to analyze user activity.

### query_users
Executes a targeted query on specific user accounts to check their current status and attributes.

### set_user_properties
Updates or changes profile data for a specified distinct user ID.

### track_custom_event
Sends a single, standalone event to the analytics platform for rapid testing purposes.

## Prompt Examples

**Prompt:** 
```
Show me a summary of our ThinkingData project configuration.
```

**Response:** 
```
I've retrieved the project summary. Your project 'Global_Quest_01' is currently active on the China-East data center. There are 45 unique events defined and 12 active cohorts. Would you like me to list the most frequent events?
```

**Prompt:** 
```
List all defined events in ThinkingData.
```

**Response:** 
```
I've retrieved the event metadata. You have events like 'UserLogin', 'LevelUp', 'PurchaseStart', and 'TutorialComplete'. Should I show you the properties associated with the 'PurchaseStart' event?
```

**Prompt:** 
```
Check the profile for user 'USER_88210934'.
```

**Response:** 
```
I've queried the user profile. USER_88210934 has a total level of 42, last logged in 2 hours ago, and belongs to the 'VIP_Gold' cohort. Would you like me to update any of their properties?
```

## Capabilities

### Get project overviews
Retrieve a high-level summary of the entire data project configuration.

### List all defined events
See metadata for every event that has been tracked within your gaming project.

### Execute deep behavioral queries
Run complex analytical queries against the collected event data to understand user funnels and actions.

### Check specific user profiles
Pull current attributes from a defined user ID, allowing you to audit accounts.

### Update user properties
Modify or set new data points on an existing user's profile for testing or correction.

### Track custom events
Send a single, specialized event to the platform when you need rapid testing or data input.

## Use Cases

### Investigating a sudden drop in engagement
A Live-Ops Manager notices purchases are down. They ask their agent, 'What events happened between login and purchase?' The agent uses `query_events` to pull the exact sequence data, showing that players are dropping off at the tutorial completion screen.

### Validating a new feature's impact
A Developer needs to test if a new onboarding step is being tracked correctly. They use `track_custom_event` via their agent, simulating 10 user completions and confirming the data appears instantly for review.

### Auditing an influential user account
A Data Analyst suspects a key tester's profile is corrupted. They ask the agent to check the profile using `query_users`, immediately seeing that attributes like 'total level' are missing, and they can use `set_user_properties` to fix it.

### Comparing different player groups
A Product Manager wants to know if VIP users behave differently than new users. They ask the agent to list user cohorts via `list_project_cohorts` and then run comparative queries on each group.

## Benefits

- Analyze complex player flows without writing SQL. Instead of building a query, you simply ask your agent to 'Show me the path from login to purchase,' letting the `query_events` tool handle the heavy lifting.
- Keep track of custom data changes easily. If you need to test how a new feature affects user profile data, use the `track_custom_event` or `set_user_properties` tools directly through your conversation.
- Get immediate status checks. Don't waste time opening dashboards; ask for an overview using `get_project_summary` to confirm which events and cohorts are active before starting a deep dive.
- Understand user groupings instantly. Instead of running multiple reports, use the agent to list defined user groups with `list_project_cohorts`, giving you immediate segmentation data.
- Audit specific users fast. Need to verify if 'USER_123' has the correct attributes? Use the `query_users` tool to pull their profile details in seconds.

## How It Works

The bottom line is you stop writing data queries and start asking questions in plain English.

1. Subscribe to this MCP and enter your required credentials: TA Project APPID, Data Token, Query Secret, and API URL.
2. Connect your preferred AI client (like Cursor or Claude).
3. Ask your agent a question like, 'Show me the user journey for players who bought Item X,' and get accurate results instantly.

## Frequently Asked Questions

**How do I check my project status using ThinkingData / 数数科技 MCP?**
Run `get_project_summary` to get an immediate overview of your entire data project. This confirms the active data center, number of events, and overall connectivity health.

**Can I find out what behavioral patterns happened in my game?**
Yes. Use `query_events` with natural language prompts to execute complex queries against historical activity data, mapping out player funnels and drop-off points.

**How do I update a user's profile using ThinkingData / 数数科技 MCP?**
You use the `set_user_properties` tool. You just tell your agent which distinct ID needs updating and what new attribute value it should receive.

**What is the difference between listing events and querying them?**
Listing events (`list_defined_events`) only gives you the names and schemas of available event types. Querying events (`query_events`) actually runs a deep query to find data instances based on those defined events.

**Does ThinkingData / 数数科技 MCP help with cohort analysis?**
Yes, you can list predefined user groups using `list_project_cohorts` and then use the agent to run comparative queries across different segments of users.