When I was testing the playwright task, it seemed that MCP was not called

[LuoPengcheng12138]

When testing the playwright task, I did not see logs of AI calling tools as I did when testing the filesystem system

python -m pipeline --exp-name birth_of_arvinxu --mcp playwright --tasks playwright/web_search/birth_of_arvinxu --models deepseek-chat --k 1
MCPMark Evaluation
Experiment: birth_of_arvinxu | 1 Model(s): deepseek-chat

============================================================
Starting evaluation 1/1: deepseek-chat
============================================================

Playwright state manager initialized
Discovering tasks in category: web_search
Discovering tasks in category: eval_web
Discovered 4 Playwright tasks across all categories

============================================================
Results Summary
============================================================
✓ Tasks passed: 0/0 (0.0%)
⏱ Total time: 0.0s
📁 Results: results/birth_of_arvinxu/deepseek-chat__playwright/run-1

============================================================
✓ All evaluations completed for 1 model(s)
============================================================
All browser resources closed

[LuoPengcheng12138]

It was I who got the Bash wrong. The correct Bash is：
python -m pipeline --exp-name birth_of_arvinxu --mcp playwright --tasks web_search/birth_of_arvinxu --models deepseek-chat --k 1