Filesystem

This guide walks you through preparing your filesystem environment for MCPMark.

1 · Configure Environment Variables

Set the FILESYSTEM_TEST_ROOT environment variable in your .mcp_env file:

Plaintext

## Filesystem
FILESYSTEM_TEST_ROOT=./test_environments

Recommended: Use FILESYSTEM_TEST_ROOT=./test_environments (relative to project root)

2 · Automatic Test Environment Download

Our code automatically downloads test folders to your specified FILESYSTEM_TEST_ROOT directory when the pipeline starts running.

Downloaded Structure:

Plaintext

./test_environments/
├── desktop/               # Desktop environment 
├── desktop_template/      # Template files for desktop
├── file_context/          # File content understanding tasks
├── file_property/         # File metadata and properties related tasks
├── folder_structure/      # Directory organization tasks
├── legal_document/        # Legal document processing
├── papers/                # Academic paper tasks
├── student_database/      # Database management tasks
├── threestudio/           # 3D Generation codebase
└── votenet/               # 3D Object Detection codebase

3 · Running Filesystem Tasks

Basic Command:

Shell

python -m pipeline --exp-name EXPNAME --mcp filesystem --tasks FILESYSTEMTASK --models MODEL --k K

Docker Usage (Recommended)

Docker is recommended to avoid library version conflicts:

Shell

# Build Docker image
./build-docker.sh

# Run with Docker
./run-task.sh --mcp filesystem --models MODEL --exp-name EXPNAME --tasks FILESYSTEMTASK --k K

Here EXPNAME refers to customized experiment name, FILESYSTEMTASK refers to the github task or task group selected (see Task Page for specific task information), MODEL refers to the selected model (see Introduction Page for model supported), K refers to the time of independent experiments.

5 · Troubleshooting

Common Issues:

• Test Environment Not Found: Ensure FILESYSTEM_TEST_ROOT is set correctly
• Prerequisites: Make sure your terminal has wget and unzip commands available
• Recommended: Use Docker to prevent library version conflicts