---
title: AI Workflow Agent M1
emoji: 🤖
colorFrom: blue
colorTo: purple
sdk: streamlit
sdk_version: "1.31.0"
python_version: "3.11"
app_file: app.py
pinned: false
---

# AI Workflow Agent - M1 (Milestone 1)

🤖 **Core AI Decision Making System** - Analyzes automation needs and generates intelligent workflows.

## Features

- 🧠 **AI Analysis**: Uses Gemini 2.5 Flash to understand your automation requirements
- ⚙️ **Workflow Generation**: Creates n8n and ComfyUI workflows from natural language
- 🔍 **Smart Search**: GitHub repository search and web search (DuckDuckGo)
- 💬 **AI Chat**: Conversational assistant for workflow planning
- ⚡ **Dual Mode**: Fast keyword matching or intelligent LLM analysis
- 🌐 **Bilingual**: Full English + Spanish support

## How to Use

1. **Enable AI Mode**: Toggle "🧠 Intelligent AI Mode" in sidebar for Gemini analysis
2. **Choose Language**: Select English or Spanish
3. **Navigate Tabs**: 
   - 🏠 **Home**: Overview and getting started
   - 🧠 **AI Analysis**: Analyze automation needs
   - ⚙️ **Build Workflow**: Generate n8n/ComfyUI JSON
   - 🔍 **GitHub Search**: Find open source projects
   - 🌐 **Web Search**: Search with DuckDuckGo
   - 💬 **Chat**: Conversational AI assistant
4. **Try Examples**: Each tab has pre-filled examples

## Technology Stack

- **LLM**: Google Gemini 2.5 Flash (REST API)
- **Workflow Tools**: n8n (automation) + ComfyUI (AI image generation)
- **Search**: DuckDuckGo (no API key) + GitHub API
- **Framework**: Streamlit 1.31.0
- **Python**: 3.11

## Configuration

**Required for AI Mode:**
- `GEMINI_API_KEY`: Get free key from [Google AI Studio](https://aistudio.google.com/app/apikeys)

**Optional:**
- `GITHUB_TOKEN`: For higher GitHub API rate limits

### Setting Secrets in HuggingFace Spaces:
1. Go to Space Settings → Variables and secrets
2. Add secret: `GEMINI_API_KEY` = your_key_here
3. Restart the Space

## Fast vs Intelligent Mode

- **⚡ Fast Mode** (AI disabled): <100ms, uses keyword pattern matching, perfect for testing
- **🧠 Intelligent Mode** (AI enabled): 2-3s, uses Gemini for deep analysis and complex workflow generation

## Example Workflows

**Customer Support Automation:**
Generate workflows that receive tickets, classify urgency, check knowledge base, create Jira tickets, and send notifications.

**Image Generation:**
Create ComfyUI workflows for txt2img, img2img, upscaling, inpainting, and ControlNet.

## Notes

- **No Docker Required**: Runs entirely on HuggingFace infrastructure
- **No Training**: Uses pre-trained Gemini model (no GPU needed)
- **Minimal Dependencies**: Only 8 packages for fast startup
- **Free Tier**: Works with free Gemini API (15 requests/minute)

---

Built for **Milestone 1** of the AI Workflow Agent project.