Files
superset2/tests/integration_tests/mcp_service/README_mcp_tests.md
Amin Ghadersohi a9d543b6f4 update: unify FastMCP server, modularize tools, and document new DAO-based architecture
- Replace dual Flask/FastAPI setup with a single, unified FastMCP server (`server.py`)
- Introduce `MCPDAOWrapper` for secure, context-aware DAO access (`dao_wrapper.py`)
- Refactor all MCP tools to be modular and domain-organized (`tools/dashboard/`, `tools/chart/`, `tools/dataset/`, `tools/system/`)
- Strongly type all tool contracts with Pydantic v2 models, including full field documentation for LLM/OpenAPI compatibility
- Refactor and extend `BaseDAO` for robust, generic CRUD/list operations
- Add and update documentation:
  - Architecture and flow diagrams (`README_ARCHITECTURE.md`)
  - Tool schema reference and usage instructions (`README.md`, `README_SCHEMAS.md`)
  - Phase 1 status and roadmap (`README_PHASE1_STATUS.md`)
- Implement and test all core list/info tools for dashboards, datasets, and charts, with full search and filter support
- Add chart creation tool (`create_chart_simple`)
- Provide extension points for Preset-specific auth, RBAC, and logging (stubbed in Phase 1)
- Prepare for LLM/agent workflows and future command-based mutations (create/update/delete)
- Expand and update unit/integration test coverage for all tools
2025-07-30 14:20:37 -04:00

932 B

Superset MCP Integration Tests

This directory contains integration tests for the Superset Model Context Protocol (MCP) service. These tests exercise all FastMCP tools for dashboards, charts, and instance metadata, ensuring correct behavior and robust schema contracts.

Contents

  • run_mcp_tests.py: Main integration test runner for all MCP tools
  • test_get_dashboard_info.py: Tests for the get_dashboard_info tool
  • test_get_dashboard_list_tools.py: Tests for dashboard listing tools

Usage

  • Ensure the MCP service is running (see superset/mcp_service/README.md)
  • Run tests with: python run_mcp_tests.py

Coverage

  • All listing, info, and filter tools are covered
  • Tests validate both Pydantic and dict responses
  • See the main MCP README and architecture docs for tool and flow details

For more on the MCP service, see superset/mcp_service/README.md and superset/mcp_service/README_ARCHITECTURE.md.