help

name	help
description	Get help with FiftyOne skills, understand available workflows, and troubleshoot setup issues

FiftyOne Skills Help

This plugin provides expert workflows for building high-quality datasets and computer vision models using FiftyOne.

When to Use This Help

• First time using FiftyOne skills
• Encountering setup or configuration issues
• Want to understand what skills are available

Prerequisites

Before using FiftyOne skills, ensure the MCP server is installed:

pip install fiftyone-mcp-server

The plugin automatically configures the MCP server connection via .mcp.json. If you installed FiftyOne in a virtual environment or conda environment, update the command path in .mcp.json to point to your environment's executable:

{
  "mcpServers": {
    "fiftyone": {
      "command": "/path/to/your/venv/bin/fiftyone-mcp"
    }
  }
}

Available Skills

FiftyOne Use

Skills for working with datasets and models:

Skill	Command	Use When
Dataset Import	/fiftyone:fiftyone-dataset-import	Importing any dataset (COCO, YOLO, VOC, videos, point clouds, multimodal, Hugging Face Hub)
Dataset Export	/fiftyone:fiftyone-dataset-export	Exporting datasets to standard formats or Hugging Face Hub for training or sharing
Find Duplicates	/fiftyone:fiftyone-find-duplicates	Removing duplicate or near-duplicate images from datasets
Dataset Inference	/fiftyone:fiftyone-dataset-inference	Running detection, classification, segmentation, or embeddings on data
Model Evaluation	/fiftyone:fiftyone-model-evaluation	Computing mAP, precision, recall, confusion matrices
Embeddings Visualization	/fiftyone:fiftyone-embeddings-visualization	Exploring dataset structure, finding clusters, identifying outliers
Create Notebook	/fiftyone:fiftyone-create-notebook	Creating Jupyter notebooks for tutorials, getting-started guides, recipes, or ML pipelines

FiftyOne Develop

Skills for developers building with FiftyOne:

Skill	Command	Use When
Develop Plugin	/fiftyone:fiftyone-develop-plugin	Creating custom operators or panels for FiftyOne App
VOODO Design	/fiftyone:fiftyone-voodo-design	Building UIs with VOODO components, styling FiftyOne panels
Code Style	/fiftyone:fiftyone-code-style	Writing Python code following FiftyOne conventions
Issue Triage	/fiftyone:fiftyone-issue-triage	Triaging GitHub issues in the FiftyOne repository

Troubleshooting Checklist

If skills aren't working as expected:

1. Verify MCP server installation

   pip show fiftyone-mcp-server
   

2. Test MCP connection Ask: "List my FiftyOne datasets"

   If this works, MCP is properly configured.

3. Check FiftyOne installation

   pip show fiftyone
   

4. Verify Python environment Ensure you're using the same environment where FiftyOne is installed.

Quick Commands

• /fiftyone:quickstart - Guided workflow to get started
• /fiftyone:help - This help document

Resources

• FiftyOne Documentation
• FiftyOne MCP Server
• FiftyOne Plugins
• Discord Community
• GitHub Issues