GuanCLI and GuanMCP Comparison
Who Should Read This Chapter
This chapter helps BI users, data analysts, IT administrators, and AI platform owners decide whether to use GuanCLI or GuanMCP first when an enterprise wants AI to use Guandata BI capabilities.
If you are not sure about the difference, remember this:
GuanCLI is a BI toolbox for AI Agents and technical users. GuanMCP is a BI connector that exposes BI query capabilities to existing enterprise AI assistants.
Core Differences
Both GuanCLI and GuanMCP let AI use Guandata BI capabilities, but their positioning and boundaries are different.
- GuanCLI is a BI toolbox for AI Agents and technical users. It requires an environment that can install and run local commands, but it has a complete Skill ecosystem and supports resource search, data query, metric query, dashboard creation, ETL creation, and more. It also supports a broader range of BI versions.
- GuanMCP is a BI connector for enterprise AI assistants. It wraps Guandata BI query capabilities as standard MCP tools so existing AI assistants can connect to BI. It focuses on read-only tasks such as querying resources, data, and metrics. It is not an entry point for creating, modifying, or publishing resources.
Simple decision rules:
- If the environment can install and run GuanCLI, prefer GuanCLI.
- Use GuanMCP only when the enterprise AI assistant cannot install or call GuanCLI and can only access external tools through MCP.
- They can be used together: GuanCLI supports analysis, delivery, O&M automation, and resource construction; GuanMCP supports read-only BI queries inside existing AI assistants.
Relationship with Guandata BI
GuanCLI and GuanMCP do not build a new data system. They reuse existing Guandata BI assets and permission systems:
- Data assets: datasets, metrics, metric trees, dashboards, cards, ETL, ChatBI topics, and more.
- Permission system: user identity, organization permissions, workspace permissions, data permissions, and resource access permissions.
- Management capabilities: resource catalogs, task status, operation records, and security audit.
The clearer the BI assets, metric definitions, and permissions are, the better AI can use BI. Before integration, sort out pilot departments, core metrics, common dashboards, and test-account permissions.
Use Case Comparison
| Item | GuanCLI | GuanMCP |
|---|---|---|
| Primary positioning | BI toolbox for AI Agents or technical teams | BI query connector for enterprise AI assistants |
| Entry point | Terminal, scripts, local Agent, delivery environment | Feishu Aily, DecideX, self-developed enterprise AI assistants |
| Typical users | AI engineers, implementation, O&M, solution teams | Business users and enterprise AI platform users |
| Integration method | Run local commands | Call through MCP tools or HTTP |
| Parameter expression | Command parameters, script parameters, organized by Agent | Schema-based parameters validated and orchestrated by the platform |
| Common tasks | Search resources, query data and metrics, troubleshoot, manage forms, build dashboards, build ETL | Read-only operations such as resource, data, and metric queries |
| Resource writes | Suitable for some creation, modification, and publishing tasks | Not used for resource creation, modification, saving, or publishing |
| Deployment dependency | Requires a local or server environment that can install and run CLI | Requires MCP Server or HTTP service integrated with the AI platform |
| BI version coverage | Broader coverage and recommended as the default option | Constrained by MCP Server capabilities and target BI version |
Capability Boundaries
What GuanCLI Is Better For
GuanCLI is suitable when the task needs to work with BI assets directly:
- Query BI resources, datasets, pages, cards, and task status.
- Preview data, troubleshoot data lineage, and locate exceptions.
- Let Agents call BI APIs through Skills.
- Operate form-entry data.
- Call ChatBI.
- Generate dashboards with GuanVis.
- Modify or debug ETL, pages, and other resources.
For delivery, implementation, or O&M automation in a controlled environment, GuanCLI is usually more direct.
What GuanMCP Is Better For
GuanMCP is suitable for registering BI query capabilities into enterprise AI platforms:
- Let business users query BI data in existing AI assistants.
- Search cards, dashboards, datasets, and ETL assets.
- Read dataset fields, card sources, and dashboard structures.
- Query metric definitions, metric data, and metric-tree attribution results.
- Call ChatBI-style Q&A capabilities.
- Return BI results according to the current user's permissions.
GuanMCP currently focuses on read-only query capabilities. It does not create, edit, save, or publish BI resources, and it does not bypass Guandata BI permissions.
How to Choose
| Customer Need | Recommended Option | Reason |
|---|---|---|
| The environment supports installing and running GuanCLI | GuanCLI | More complete capabilities, Skill ecosystem, and broader BI version support. |
| Users want to ask business data questions directly in Feishu Aily or another AI assistant, but GuanCLI cannot be installed or called | GuanMCP | GuanMCP can act as a read-only BI connector when tools must be integrated through MCP. |
| AI helps implementation users troubleshoot datasets, pages, ETL, or tasks | GuanCLI | Local commands and Skills better support controlled troubleshooting. |
| AI automatically drafts dashboards | GuanCLI + GuanVis | This requires reading dataset structure and publishing pages, which is resource generation. |
| The enterprise has a unified AI platform and wants BI as a query tool, but cannot install or call GuanCLI | GuanMCP | MCP provides schema-based query tools for platform orchestration. |
| The customer has no AI assistant and only wants AI tools to access BI first | GuanCLI | CLI is easier to validate from a local environment. |
| Existing BI assets and AI assistant are both available, and business users need low-cost Q&A but cannot run GuanCLI | GuanMCP | Reuses BI data, metrics, pages, and permissions for read-only connection. |
| Batch manage, migrate, or modify BI assets | GuanCLI or BI product UI | GuanMCP is not a resource management entry. |
Reading Guide by Role
| Role | Start With | Focus |
|---|---|---|
| Business analyst | GuanMCP User Guide | What can be asked in AI assistants, where answers come from, and how permissions work. |
| IT administrator / platform administrator | GuanMCP Deployment Guide | Service deployment, entry authentication, user identity passing, and permission validation. |
| Data analyst / implementation consultant | GuanCLI User Guide | Query resources, preview data, troubleshoot tasks, and call ChatBI. |
| BI developer | GuanVis User Guide, GuanETL User Guide | Generate dashboards, modify ETL, check results, and confirm before saving. |
Recommended Combination
In mature environments, GuanCLI and GuanMCP can be used together.
GuanMCP serves the business side by connecting trusted BI query capabilities to existing AI assistants. Business users keep asking questions from their usual work entry, while answers still come from BI and BI permissions still apply.
GuanCLI serves the construction side by helping implementation, O&M, and technical teams analyze, troubleshoot, generate, and configure resources. It is suitable for test environments, delivery environments, and controlled automation workflows.
Pilot Suggestions
For first integration, start with a clearly scoped business subject such as sales operations, store operations, or financial analysis:
- Select a group of datasets, metrics, and dashboards that are already used stably in BI.
- Prepare three test accounts with high, normal, and low permissions.
- Use GuanCLI to check whether resources are searchable and whether field and metric definitions are clear.
- Connect GuanMCP to the AI assistant and verify whether business users can ask correct questions in the original entry.
- Validate permissions, definitions, returned content, and audit records before expanding to more departments.