Course 2: Building with MIFY — Chapter 5

Chapter 5: Advanced Patterns

Build complex solutions using MIFY’s advanced features.

What You’ll Learn

• Batch processing for scale
• Sub-workflows for modularity
• Agent orchestration
• Model training

Batch Processing

Process many items in parallel:

# Create a batch job
curl -X POST \
  -H "Authorization: Bearer mify_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "graphId": "your-graph-id",
    "items": [{"input": "item1"}, {"input": "item2"}, {"input": "item3"}],
    "concurrency": 3
  }' \
  https://your-instance/api/batch

Control execution:

POST /api/batch/{id}/start    # Start batch
POST /api/batch/{id}/pause    # Pause
POST /api/batch/{id}/resume   # Resume
POST /api/batch/{id}/cancel   # Cancel

Sub-Workflows

Use the Sub-Workflow node to call one workflow from another:

• Build reusable components (e.g., “email validation” workflow)
• Call them from multiple parent workflows
• Keep complex workflows modular

Agent Nodes

MIFY has 6 agent nodes:

Node	What It Does
ReAct Agent	Reasoning + Acting loop with tools
Multi-Agent	Coordinate multiple agents
CLI Agent	Run coding agents (Claude Code, Codex)
EnvInspect	Check environment (OS, tools, versions)
Browser Use	Automate browser interactions
Computer Use	Control desktop applications

Model Training

Train custom models via the execution layer:

# 1. Create a dataset
POST /api/execution/training/datasets/upload

# 2. Upload files
POST /api/execution/training/datasets/{id}/presign-file
# Upload to the presigned R2 URL

# 3. Finalize
POST /api/execution/training/datasets/{id}/finalize

# 4. Submit training
POST /api/execution/training/submit

Monitor at /training.

Platform Integrations (Hub)

Import and connect to 28+ platforms via the Integration Hub at /integrations:

• Native adapters: Flowise, Dify, Langflow, n8n, OpenClaw, RAGFlow, DocuSeal
• Registry-driven: ActivePieces, Node-RED, Trigger.dev, Sim Studio, and more

Course 2 Knowledge Check

1. How do you authenticate API calls? → Bearer token with mify_ API key
2. What are the three content generation paths? → MIFY Queue (async GPU), AI Provider (sync/stream), ComfyUI Local
3. How do you create a plugin? → npx mify-plugin init, implement node, npx mify-plugin publish
4. What is RAG? → Retrieve relevant documents, then generate answers using that context
5. How do you run batch jobs? → POST /api/batch with items array and concurrency setting

Previous: Chapter 4 — Creating a Plugin | Back to Courses