FAQ

We’ve compiled frequently asked questions in advance.
If you have any questions, please feel free to contact us at aihub@querypie.com.

Basic Usage

Our company uses QueryPie. How does it connect with AI Hub?

Thank you for using QueryPie! 🙇‍♀️
QueryPie AI Hub provides a QueryPie Client MCP server that corresponds to the DAC/SAC user console.
Instead of directly accessing internal resources through QueryPie, you can make natural language requests to the LLM through AI Hub and receive results.
For detailed guidance, check the QueryPie Client MCP Server documentation.

What models are currently available?

The currently available model is Claude 4 Sonnet.
We are actively testing internally to provide other models 🏃‍♀️‍➡️

Can I use my own LLM API?

We currently do not provide user LLM API integration functionality.

How much can I use?

During the CBT period, you can call the LLM without usage limits.
We appreciate your testing and feedback. 🙏

Tool calls are not performed more than 12 times.

By default, the recursion limit is set to 25.
In this case, tool calls and responses are each counted as 1, so actual tool calls cannot exceed 12 times.
You can increase the recursion limit up to 100 in the Settings modal.

What MCP servers are currently available?

We provide carefully selected MCP servers that can be immediately applied to daily work.

AWS
n8n
Dify
Slack
Google Calendar, Gmail, Google Sheets
Jira, Confluence

We also provide various other MCP servers. Check the Available MCP Servers documentation.

What MCP servers will be added in the future?

Salesforce
ServiceNow
LlamaIndex

Various other MCP servers will continue to be added. If you want to know what MCP servers are planned to be added, check the Roadmap documentation.

The MCP server I want is not on the list!

What MCP server do you need? Please feel free to request it at aihub@querypie.com!

Permission Management

How much control can users have over permissions for connected services?

Currently, we inherit the permission scope allowed by the authentication information (API Key, user’s OAuth token, etc.) used for Integration addition.

For example, when AWS is integrated with AI Hub:

The actions available in AI Hub are determined by which Access Key you specified in AI Hub.
If the Access Key is linked to an administrator account, AI Hub may be able to execute prompts appropriately.
However, AI Hub cannot execute functions that the MCP server does not support.

When the Access Control feature is released in the future, you will be able to set call permissions for specific MCP Tools based on attributes of users, MCP Tools, Agents, etc.

Custom MCP Server

I added a custom MCP server. Who else can use it besides me?

Custom MCP servers can only be used by the user who added them.
Other users in the same company cannot see or use that custom MCP server. Custom MCP servers added through user-side integration are considered personal settings for that user.

Security

How is the security of MCP servers provided by QueryPie guaranteed?

The servers we provide are directly implemented using secure libraries.
We also perform internal penetration testing and vulnerability assessment processes as follows:

Infrastructure and Configuration Review

AWS environment review
Network configuration review
Access control configuration

Security Review

Image vulnerability assessment
EC2 infrastructure vulnerability assessment
Penetration testing
SCA (Software Composition Analysis)
SAST (Static Application Security Testing)
DAST (Dynamic Application Security Testing)

FAQ