Biomni Dify Plugin - Complete Setup Guide

📁 Project Structure

Create this exact folder structure:

biomni_plugin/
├── manifest.yaml                # Plugin manifest
├── main.py                      # Entry point
├── requirements.txt             # Dependencies
├── .env.example                 # Environment template
├── _assets/
│   └── biomni_icon.png         # Plugin icon (add your icon here)
├── provider/
│   ├── __init__.py             # Empty file
│   ├── biomni.yaml             # Provider config
│   └── biomni.py               # Provider implementation
└── tools/
    ├── __init__.py             # Empty file
    ├── biomni_agent.yaml       # Tool config
    └── biomni_agent.py         # Tool implementation

🔧 Configuration Steps

Step 1: Update Plugin Metadata

Ensure metadata is correct in these files (author, labels, etc.):

• provider/biomni.yaml
• tools/biomni_agent.yaml
• manifest.yaml

Step 2: Add Your Icon

• Create or find a PNG icon for Biomni (recommended size: 256x256px)
• Place it at _assets/biomni_icon.png

Step 3: Configure Biomni Agent Import (via environment variables)

Set these environment variables to point the plugin at your Biomni agent:

# Optional extra sys.path entry if Biomni lives outside site-packages
export BIOMNI_PYTHON_PATH=/absolute/path/to/biomni/source

# Import path to your agent class or instance. Examples:
#   Package class:  biomni:BiomniAgent
#   Module object:  biomni.agent:agent
export BIOMNI_AGENT_IMPORT="biomni:BiomniAgent"

# Optional method name to call on the agent (defaults to 'go').
export BIOMNI_AGENT_METHOD=go

Step 4: Agent Method Call

tools/biomni_agent.py will call BIOMNI_AGENT_METHOD if set, otherwise tries go, run, process_query, then __call__.

Step 5: Configure Result Formatting (Optional)

In tools/biomni_agent.py, customize the _format_result() method based on your agent's output format.

Step 6: Update Dependencies

In requirements.txt, add all dependencies your Biomni agent needs. If Biomni is pip-installable, uncomment and pin biomni.

Step 7: Create Empty init.py Files

Create empty __init__.py files:

touch provider/__init__.py
touch tools/__init__.py

🚀 Development and Testing

Step 1: Install Dify CLI Tool

Download the Dify plugin CLI tool:

# Download from Dify's GitHub releases or documentation
# Follow Dify's official installation guide for the CLI tool

Step 2: Set Up Development Environment

cd biomni_plugin
cp .env.example .env

# Get remote debugging credentials from your Dify instance:
# 1. Go to Dify admin panel → Plugin Management
# 2. Click "Get Debugging Key" 
# 3. Copy the host, port, and key values
# 4. Update your .env file with these values

Step 3: Install Dependencies

pip install -r requirements.txt

Step 4: Test the Plugin Locally

# Run in development mode
python main.py

# Or use the CLI tool for debugging
dify plugin dev

Step 5: Remote Debugging

With your .env file configured:

python -m main

You should see the plugin appear in your Dify workspace for testing.

📦 Packaging the Plugin

Method 1: Using Dify CLI Tool (Recommended)

# In your plugin root directory
dify plugin package .

# This creates biomni.difypkg file

Method 2: Manual Packaging

If CLI tool isn't available:

# Create a zip file with all plugin contents
zip -r biomni.difypkg \
    manifest.yaml \
    main.py \
    requirements.txt \
    _assets/ \
    provider/ \
    tools/

# Note: Don't include .env, __pycache__, or other development files

🔍 Testing Your Plugin

Unit Testing

Create a simple test script:

# test_plugin.py
import sys
sys.path.append('/path/to/your/biomni')

def test_agent_import():
    """Test if your agent can be imported"""
    try:
        # Use your actual import
        from your_biomni_module import agent
        print("✅ Agent import successful")
        return True
    except Exception as e:
        print(f"❌ Agent import failed: {e}")
        return False

def test_agent_call():
    """Test if your agent method works"""
    try:
        from your_biomni_module import agent
        result = agent.go("test query")
        print(f"✅ Agent call successful: {result[:100]}...")
        return True
    except Exception as e:
        print(f"❌ Agent call failed: {e}")
        return False

if __name__ == "__main__":
    test_agent_import()
    test_agent_call()

Integration Testing

1. Install the plugin in your Dify development environment
2. Create a simple Agent or Workflow
3. Add your Biomni tool
4. Test with simple queries first
5. Gradually test more complex biomedical queries

🚀 Deployment

Development Environment

# Upload to your Dify development instance
# Use the plugin management interface to upload biomni.difypkg

Production Environment

# For production, ensure:
# 1. All dependencies are properly configured
# 2. Biomni agent is installed and accessible
# 3. Sufficient system resources (memory, CPU)
# 4. Proper error handling and logging

🔧 Troubleshooting

Common Issues:

1. Import Errors: Check your Biomni path and import statements
2. Agent Not Found: Verify your agent setup and method names
3. Timeout Issues: Increase max_execution_time for complex queries
4. Memory Issues: Monitor resource usage, add cleanup if needed
5. Permission Errors: Ensure proper file permissions and paths

Debug Logging:

Add more logging to troubleshoot:

import logging
logging.basicConfig(level=logging.DEBUG)
logger = logging.getLogger(__name__)

# Add debug logs throughout your code
logger.debug(f"Agent setup status: {self.agent is not None}")
logger.debug(f"Processing query: {research_query[:50]}...")

✅ Final Checklist

Before packaging:

• Updated all your-name placeholders
• Added proper Biomni agent import and path
• Configured correct agent method call
• Added plugin icon to _assets/
• Updated requirements.txt with dependencies
• Created empty __init__.py files
• Tested agent import and method calls
• Tested plugin in development mode
• Verified remote debugging works

📚 Next Steps

After successful deployment:

1. Monitor plugin performance and resource usage
2. Collect user feedback on biomedical research capabilities
3. Consider adding more advanced features:
   • Progress streaming for long-running analyses
   • Result caching for common queries
   • Integration with biomedical databases
   • Batch processing capabilities
4. Submit to Dify Marketplace (optional)

Your Biomni plugin is now ready for use! 🧬✨