Menu
Create, query, and manage MULLER datasets (multimodal data lake with version control). Use when user wants to work with datasets, create tensors, append data, query samples, or inspect dataset information.
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
Advanced query operations for MULLER datasets - indexing, vector search, aggregation, and complex filtering. Use when user wants to create indexes, perform vector similarity search, or run aggregation queries.
Export and integrate MULLER datasets with other formats - Arrow, Parquet, JSON, NumPy, MindRecord. Use when user wants to export data, convert formats, or integrate with other frameworks.
Git-like version control for MULLER datasets - commit, branch, merge, diff, and conflict resolution. Use when user wants to version datasets, create branches, merge changes, or view history.
| name | muller-dataset |
| description | Create, query, and manage MULLER datasets (multimodal data lake with version control). Use when user wants to work with datasets, create tensors, append data, query samples, or inspect dataset information. |
| compatibility | Requires Python 3.11+, muller package installed |
DO NOT create new Python files. Always use the existing scripts provided in this skill:
scripts/dataset_manager.py for dataset and tensor managementscripts/data_operations.py for data CRUD operationsExecute these scripts directly with python3 command. Never write new scripts to the project root.
When executing operations:
Use this skill when the user wants to:
Manages dataset lifecycle and structure.
Operations:
create - Create a new datasetload - Load existing dataset infodelete - Delete a datasetinfo - Get dataset informationstats - Get dataset statisticscreate-tensor - Create a new tensordelete-tensor - Delete a tensorrename-tensor - Rename a tensorUsage:
# Create dataset
python3 .claude/skills/muller-dataset/scripts/dataset_manager.py create --path ./my_dataset
# Create with tensors
python3 .claude/skills/muller-dataset/scripts/dataset_manager.py create --path ./my_dataset \
--tensors "images:image:jpg,labels:class_label:uint32"
# Get info
python3 .claude/skills/muller-dataset/scripts/dataset_manager.py info --path ./my_dataset
# Create tensor
python3 .claude/skills/muller-dataset/scripts/dataset_manager.py create-tensor --path ./my_dataset \
--name embeddings --htype vector --dtype float32
Handles data CRUD operations.
Operations:
append - Add single sampleextend - Add multiple samplesupdate - Update existing sampledelete - Delete samplesquery - Query and filter samplesimport - Import data from filesUsage:
# Append sample
python3 .claude/skills/muller-dataset/scripts/data_operations.py append --path ./my_dataset \
--data '{"images": "path/to/img.jpg", "labels": 1}'
# Query samples
python3 .claude/skills/muller-dataset/scripts/data_operations.py query --path ./my_dataset \
--filter "labels > 5" --limit 10
# Import from file
python3 .claude/skills/muller-dataset/scripts/data_operations.py import --path ./my_dataset \
--source data.jsonl
For detailed workflows and examples, check references/ directory.